US 6,159,014 AGrant

Method and Apparatus for Training of Cognitive and Memory Systems in Humans

Issue Date:2000-12-12

•19 Claims

•24 Drawing Sheets

Abstract

An apparatus and method for training the cognitive and memory systems in a subject is provided. The apparatus and method incorporates a number of different programs to be played by the subject. The programs artificially process selected portions of language elements, called phonemes, so they will be more easily distinguished by the subject, and gradually improves the subject's neurological processing and memory of the elements through repetitive stimulation. The programs continually monitor a subject's ability to distinguish the processed language elements, and adaptively configures the programs to challenge and reward the subject by altering the degree of processing. Through adaptive control and repetition of processed speech elements, and presentation of the speech elements in a creative fashion, a subject's cognitive processing of acoustic events common to speech, and memory of language constructs associated with speech elements are significantly improved.

Metadata

Assignee

Scientific Learning Corp.

Inventors

William M. Jenkins
Michael M. Merzenich
Steven Lamont Miller
Bret E. Peterson
Paula Tallal

Application Information

Application Number:US 9920722

Filing Date:1997-12-17

Priority Date:1997-12-17

Art Unit:373

Classifications

IPC:

G09B 500

Field of Search:

434463273345704706112;116;118;156;169;185;219;234;236-238;247;258-261;307 R;308;365;323;332 ;3361-15;37237;273;460145;348243;265;270;503;50415;46

Patent Drawings (24 sheets)

Description

Microfiche Appendix

This application includes a microfiche appendix consisting of a microfiche with a total of 12 frames.

Cross Reference to Related Applications

This application is related to co-pending U.S. patent application Ser. No. 08/982,189 now entitled "METHOD AND APPARATUS FOR TRAINING OF SENSORY AND PERCEPTUAL SYSTEMS IN LLI SUBJECTS"; and U.S. patent application Ser. No. 08/992,071 filed Dec. 17, 1997 entitled "METHOD AND APPARATUS FOR TRAINING OF SENSORY AND PERCEPTUAL SYSTEMS IN LLI SUBJECTS"; both assigned to Scientific Learning Corporation, and incorporated herein by reference for all purposes.

Background of the Invention

1. Field of the Invention

This invention relates in general to the field of education in human subjects, and more specifically to a computer program for training the cognitive and memory processing systems in human subjects.

2. Description of the Related Art

Up to ten percent of humans have some type of language-learning impairments (LLI) resulting from the inability to accurately process short duration acoustic events at the rates that occur in normal speech. Their trouble distinguishing among elements of speech is neurologically based and has far reaching consequences: academic failure, emotional and disciplinary problems, and possibly diminished lifelong achievement and self-image. No bracket of intelligence, race, gender or economic level is immune from this problem.

More specifically, people with LLI have difficulty detecting and identifying sounds that occur simultaneously or in close proximity to each other--a phenomenon known as "masking." Because of masking, people with LLI require sounds that are as much as 45 decibels more intense than the preceding or subsequent masking noise to distinguish and understand them. In addition, people with LLI are consistently poorer at detecting a brief tone presented with a masking noise, particularly when the brief tone is turned on immediately prior to the masking noise. This phenomenon is called "backward masking." Similarly, when the brief tone is turned on immediately after the masking noise a similar decrease in detectability can occur. This phenomenon is called "forward masking". For a tone to be detected by a person with LLI in the presence of a masking noise, the tone must be separated in time or frequency from the masking noise.

One way individuals develop such auditory processing problems is from middle ear infections when they are young and beginning to develop the oral representations of language in the central auditory nervous system. When a child has an ear infection, fluid can build up and block or muffle the sound wave entering the ear causing intermittent hearing loss. Even if the infection doesn't permanently damage the ear, the child's brain doesn't learn to process some sounds because it hasn't heard them accurately before, on a consistent basis. This typically occurs during a critical period of brain development when the brain is building the nerve connections necessary to accurately process acoustic events associated with normal speech. As the child grows up, the nervous system never builds, or rebuilds the connections necessary to properly process normal speech.

Researchers believe that the auditory processing problem is essentially one of timing. Vowel sounds like /a/ and /e/ usually last at least 100 milliseconds and typically have constant frequency content. Consonants, on the other hand, typically have modulated frequency components, and last less than 40 milliseconds. People with LLI cannot process these faster speech elements, especially the hard consonants like /t/, /p/, /d/ and /b/, if they occur either immediately before or after vowels, or if they are located near other consonants. Rather than hearing the individual sounds that make up a particular phoneme, people with LLI integrate closely associated sounds together over time. Since the duration of vowels are typically longer than consonants, the modulated frequency portions of consonants are often lost in the integration, an affect that may also hinder the resolution of the vowel, particularly short duration vowels.

This problem of abnormal temporal integration of acoustic events over time is not limited to people with LLI. Rather, the problem extends to stroke victims who have lost the neurological connections necessary to process speech, as well as to individuals raised in one country, having one set of language phonemes, and attempting to learn the language of another country, having a distinct set of language phonemes. For example, it is known that an individual raised in Japan is not often presented with phonemes similar to the English r's and l's, because those consonants are not common in the Japanese language. Similarly, there are many subtleties in the sounds made by a speaker of Japanese that are difficult to distinguish unless raised in Japan. The phonetic differences between languages are distinctions that must be learned, and are often very difficult. But, they are clearly problems that relate to the temporal processing of short duration acoustic events.

The above described temporal processing deficiency has little if anything to do with intelligence. In fact, some LLI specialists argue that brains choosing this different route by which to absorb and reassemble bits of speech may actually stimulate creative intelligence, but at the expense of speech and reading problems.

Recent studies have shown that if the acoustic events associated with phonemes that are difficult to distinguish, such as /ba/ and /da/, are slowed down, or that the consonant portion of the phonemes are emphasized, that students diagnosed as LLI can accurately distinguish between the phonemes. In addition, if the interval between two complex sounds is lengthened, LLI students are better able to process the sounds distinctly.

Heretofore, the solution to the processing problem has been to place LLI students in extended special education and/or speech therapy training programs that focus on speech recognition and speech production. Or, more commonly, repetitive reading programs, phonic games, or other phonic programs are undertaken. These programs often last for years, with a success rate that is often more closely associated with the skill of the speech and language professional than with the program of study.

In addition to the temporal processing problems discussed above, many individuals have a decreased ability to remember and process logical constructs that are common in language. For example, a simple command such as "place the small red circle inside the large green square", requires an individual to logically associate the adjectives "small" and "red" with the subject "circle", and the adjectives "large" and "green" with the subject "square". The command further requires that the individual manipulate the first subject "circle" by placing it into the second subject "square". Individuals with abnormal cognitive processing, or with abnormal short term memory, become easily confused when presented with such commands.

What is needed is a method and apparatus that allows a subject with abnormal temporal processing to train, or retrain their brain to recognize and distinguish short duration acoustic events that are common in speech. Moreover, what is needed is a program that repetitively trains a subject to distinguish phonemes at a normal rate, by first stretching, and/or emphasizing elements of speech to the point that they are distinguishable, or separating speech elements in time, and then adaptively adjusting the stretching, emphasis and separation of the speech elements to the level of normal speech. The adaptive adjustments should be made so as to encourage the subject to continue with the repetitions, and the number of repetitions should be sufficient to develop the necessary neurological connections for normal temporal processing of speech. Moreover, the program should provide acoustic signals to the brain that are better for phonetic training than normal human speech.

Furthermore, what is needed is an adaptive training method to train individuals to accurately process logical constructs that are common in language. In addition, what is needed is a method to improve an individual's short term memory to allow them to correctly process both simple and complex language constructs.

Summary

To address the above-detailed deficiencies, the present invention provides a method to improve a subject's comprehension of logical constructs within language using modified speech commands. The method provides a plurality of training skill levels for the speech commands, the plurality of training skill levels differing from each other in the difficulty of logical constructs, and in the amount of audio processing applied to the speech commands. The method also selects from the plurality of training skill levels, a training skill level to be presented to the subject that is associated with, or corresponds to, the subject's ability. The method also presents a game board that has a plurality of graphical components, the plurality of graphical components being the subject of the modified speech commands. The method further presents as a trial, a speech command from a set of speech commands from the selected skill level. The speech command directs the subject to manipulate at least one of the plurality of graphical components. If the subject correctly manipulates the graphical components, the method presents another trial. If the subject incorrectly manipulates the graphical components, the method visually indicates to the subject the correct manipulation. As the subject correctly manipulates the graphical components, the amount of processing applied to the speech commands is reduced. And, as the subject incorrectly manipulates the graphical components, the amount of processing applied to the speech commands is increased. Thus, the method adaptively trains the subject to correctly distinguish between similar speech commands.

In another aspect, the present invention provides a method to improve the cognitive processing system of a subject. The method is presented to the subject on a computer, and interacts with the subject via input/output devices on the computer. The method provides a plurality of stimulus sets, each of the plurality of stimulus sets having a plurality of command sentences. The method also provides a plurality of target graphical images, each of the graphical images associated with a different one of the plurality of command sentences. The method further provides a plurality of distractor images that are not associated with the plurality of command sentences. The method then presents to the subject one of the plurality of command sentences from one of the plurality of stimulus sets to the subject, the presented sentence modified acoustically, and presents to the subject a target graphical image, from the plurality of target graphical images, that is associated with the presented command sentence. Along with the presented target graphical image, the method presents a plurality of distractor images. The subject is then required to distinguish between the presented target graphical image, and the presented plurality of distractor images by selecting the target graphical image associated with the presented command sentence.

In yet another aspect, the present invention provides an adaptive method to improve a subject's short term memory. The method utilizes a computer to process and present sounds to the subject. The method displays a plurality of graphical images on the computer, the graphical images associated with acoustic events. The method associates in pairs the plurality of graphical images with particular acoustic events such that two different graphical images are associated with a particular acoustic event. Upon the subject's selection of any of the plurality of graphical images, its associated acoustic event is presented. The method then requires the user to discriminate between the acoustic events by sequentially selecting two different graphical images from among the plurality of graphical images, that are associated with the particular acoustic event. The acoustic events are modified by stretching them in the time domain by varying amounts. As the subject correctly remembers the acoustic events at one skill level, the amount of stretching applied to the acoustic events is reduced. In addition, as the subject correctly remembers the acoustic events, the number of graphical image pairs presented to the subject increases, requiring the subject to better train his/her memory.

Brief Description of the Drawings

These and other objects, features, and advantages of the present invention will become better understood with regard to the following description, and accompanying drawings where:

FIG. 1 is a block diagram of a computer system for executing a program according to the present invention.

FIG. 2 is a block diagram of a computer network for executing a program according to the present invention.

FIG. 3 is a chart illustrating frequency/energy characteristics of two phonemes within the English language.

FIG. 4 is a chart illustrating auditory reception of a phoneme by a subject having normal receptive characteristics, and by a subject whose receptive processing is impaired.

FIG. 5 is a chart illustrating stretching of a frequency envelope in time, according to the present invention.

FIG. 6 is a chart illustrating emphasis of selected frequency components, according to the present invention.

FIG. 7 is a chart illustrating up-down frequency sweeps of varying duration, separated by a selectable inter-stimulus-interval (ISI), according to the present invention.

FIG. 8 is a pictorial representation of a game selection screen according to the present invention.

FIG. 9 is a pictorial representation of a game entitled "Old MacDonald's Flying Farm" according to the present invention.

FIG. 10 is a flow chart illustrating the adaptive auditory training procedures embodied in the game Old MacDonald's Flying Farm.

FIGS. 11 and 12 are pictorial representations of a game entitled "Block Commander" according to the present invention.

FIG. 13 is a flow chart illustrating the adaptive auditory training procedures embodied in the game Block Commander.

FIGS. 14 and 15 are pictorial representations of a game entitled "Circus Sequence" according to the present invention.

FIG. 16 is a flow chart illustrating the initial training procedures embodied in the game Circus Sequence.

FIG. 17 is a flow chart illustrating the adaptive auditory training procedures embodied in the game Circus Sequence.

FIG. 18 is a pictorial representation of a game entitled "Phonic Match" according to the present invention.

FIG. 19 includes two tables illustrating the processing levels and the training levels embodied in the game Phonic Match.

FIG. 20 is a flow chart illustrating the adaptive auditory training process embodied in the game Phonic Match.

FIGS. 21 and 22 are pictorial representations of a game entitled "Phonic Words" according to the present invention.

FIG. 23 is a flow chart illustrating the adaptive auditory training process embodied in the game Phonic Words.

FIGS. 24 and 25 are pictorial representations of a game entitled "Phoneme Identification" according to the present invention.

FIG. 26 is a flow chart illustrating the initial training procedures embodied in the game Phoneme Identification.

FIG. 27 is a flow chart illustrating the adaptive auditory training process embodied in the game Phoneme Identification.

FIG. 28 is a pictorial representation of a game entitled "Language Comprehension Builder" according to the present invention.

FIG. 29 is a flow chart illustrating the initial training procedures embodied in the game Language Comprehension Builder.

FIG. 30 is a flow chart illustrating the adaptive auditory training procedures embodied in the game Language Comprehension Builder.

FIG. 31 is a flow chart illustrating a time-scale modification algorithm for modifying acoustic elements according to the present invention.

FIG. 32 is a flow chart illustrating a filter-bank summation emphasis algorithm for modifying acoustic elements according to the present invention.

FIG. 33 is a flow chart illustrating an overlap-add emphasis algorithm for modifying acoustic elements according to the present invention.

Detailed Description

Referring to FIG. 1, a computer system 100 is shown for executing a computer program to train, or retrain a learning language impaired (LLI) subject, according to the present invention. The computer system 100 contains a computer 102, having a CPU, memory, hard disk and CD ROM drive (not shown), attached to a monitor 104. The monitor 104 provides visual prompting and feedback to the subject during execution of the computer program. Attached to the computer 102 are a keyboard 105, speakers 106, a mouse 108, and headphones 110. The speakers 106 and the headphones 110 provide auditory prompting and feedback to the subject during execution of the computer program. The mouse 108 allows the subject to navigate through the computer program, and to select particular responses after visual or auditory prompting by the computer program. The keyboard 105 allows an instructor to enter alpha numeric information about the subject into the computer 102. Although a number of different computer platforms are applicable to the present invention, embodiments of the present invention execute on either IBM compatible computers or Macintosh computers.

Now referring to FIG. 2, a computer network 200 is shown. The computer network 200 contains computers 202, 204, similar to that described above with reference to FIG. 1, connected to a server 206. The connection between the computers 202, 204 and the server 206 can be made via a local area network (LAN), a wide area network (WAN), or via modem connections, directly or through the Internet. A printer 208 is shown connected to the computer 202 to illustrate that a subject can print out reports associated with the computer program of the present invention. The computer network 200 allows information such as test scores, game statistics, and other subject information to flow from a subject's computer 202, 204 to a server 206. An administrator can then review the information and can then download configuration and control information pertaining to a particular subject, back to the subject's computer 202, 204. Details of the type of information passed between a subject's computer and a server are provided in co-pending U.S. application Ser. No. 08/995,680, entitled "Remote Computer-Assisted Professionally Supervised Teaching System", assigned to Scientific Learning Corporation.

Before providing a detailed description of the present invention, a brief overview of certain components of speech will be provided, along with an explanation of how these components are processed by LLI subjects. Following the overview, general information on speech processing will be provided so that the reader will better appreciate the novel aspects of the present invention.

Referring to FIG. 3, a chart is shown that illustrates frequency components, over time, for two distinct phonemes within the English language. Although different phoneme combinations are applicable to illustrate features of the present invention, the phonemes /daa/ and /ba/ are shown. For the phoneme /da/, a downward sweep frequency component 302, at approximately 2.5-2 khz is shown to occur over a 35 ms interval. In addition, a downward sweep frequency component 304, at approximately 1 khz is shown to occur during the same 35 ms interval. At the end of the 35 ms interval, a constant frequency component 306 is shown, whose duration is approximately 110 ms. Thus, in producing the phoneme /da/, the stop consonant portion of the element /d/ is generated, having high frequency sweeps of short duration, followed by a long vowel element /a/ of constant frequency.

Also shown are frequency components for a phoneme /ba/. This phoneme contains an upward sweep frequency component 308, at approximately 2 khz, having a duration of approximately 35 ms. The phoneme also contains an upward sweep frequency component 310, at approximately 1 khz, during the same 35 ms period. Following the stop consonant portion /b/ of the phoneme, is a constant frequency vowel portion 314 whose duration is approximately 110 ms.

Thus, both the /ba/ and /da/ phonemes begin with stop consonants having modulated frequency components of relatively short duration, followed by a constant frequency vowel component of longer duration. The distinction between the phonemes exist primarily in the 2 khz sweeps during the initial 35 ms interval. Similarity exists between other stop consonants such as /ta/, /pa/, /ka/ and /ga/.

Referring now to FIG. 4, the amplitude of a phoneme, for example /ba/, is viewed in the time domain. A short duration high amplitude peak waveform 402 is created upon release of either the lips or the tongue when speaking the consonant portion of the phoneme, that rapidly declines to a constant amplitude signal of longer duration. For an individual with normal temporal processing, the waveform 402 will be understood and processed essentially as it is. However, for an individual who is learning-language impaired, or who has abnormal temporal processing, the short duration, higher frequency consonant burst will be integrated over time with the lower frequency vowel, and depending on the degree of impairment, will be heard as the waveform 404. The result is that the information contained in the higher frequency sweeps associated with consonant differences, will be muddled, or indistinguishable.

With the above general background of speech elements, and how LLI subjects process them, a general overview of speech processing will now be provided. As mentioned above, one problem that exists in LLI subjects is the inability to distinguish between short duration acoustic events. If the duration of these acoustic events are stretched, in the time domain, it is possible to train LLI subjects to distinguish between these acoustic events. An example of such time domain stretching is shown in FIG. 5, to which attention is now directed.

In FIG. 5, a frequency vs. time graph 500 is shown that illustrates a waveform 502 having short duration characteristics similar to the waveform 402 described above. Using existing computer technology, the analog waveform 502 can be sampled and converted into digital values (using a Fast Fourier Transform, for example). The values can then be manipulated so as to stretch the waveform in the time domain to a predetermined length, while preserving the amplitude and frequency components of the modified waveform. The modified waveform can then be converted back into an analog waveform (using an inverse FFT) for reproduction by a computer, or by some other audio device. The waveform 502 is shown stretched in the time domain to durations of 60 ms (waveform 504), and 80 ms (waveform 506). By stretching the consonant portion of the waveform 502 without effecting its frequency components, subjects with LLI can begin to hear distinctions in common phonemes.

Another method that may be used to help LLI subjects distinguish between phonemes is to emphasize selected frequency envelopes within a phoneme. Referring to FIG. 6, a graph 600 is shown illustrating a frequency envelope 602 whose envelope varies by approximately 27 hz. By detecting frequency modulated envelopes that vary from say 3-30 hz, similar to frequency variations in the consonant portion of phonemes, and selectively emphasizing those envelopes, they are made more easily detectable by LLI subjects. A 10 dB emphasis of the envelope 602 is shown in waveform 604, and a 20 dB emphasis in the waveform 606.

A third method that may be used to train LLI subjects to distinguish short duration acoustic events is to provide frequency sweeps of varying duration, separated by a predetermined interval, as shown in FIG. 7. More specifically, an upward frequency sweep 702, and a downward frequency sweep 704 are shown, having duration's varying between 25 and 80 milliseconds, and separated by an interstimulus interval (ISI) of between 500 and 0 milliseconds. The duration and frequency of the sweeps, and the interstimulus interval between the sweeps are varied depending on the processing level of the LLI subject, as will be further described below.

Utilization of up-down frequency sweeps with varying ISI has been fully described in U.S. patent application Ser. No. 08/351,803, entitled "METHOD AND DEVICE FOR ENHANCING THE RECOGNITION OF SPEECH AMONG SPEECH-IMPAIRED INDIVIDUALS", and is hereby incorporated by reference.

Each of the above described methods have been combined in a unique fashion by the present invention to provide an adaptive training method and apparatus for training subjects having abnormal temporal processing abilities to recognize and distinguish short duration acoustic events that are common in speech. The present invention is embodied into a computer program entitled Fast ForWord by Scientific Learning Corporation. The computer program is provided to an LLI subject via a CD-ROM which is input into a general purpose computer such as that described above with reference to FIG. 1. In addition, a user may log onto a server, via an Internet connection, for example, to upload test results, and to download training parameters for future exercises. Specifics of the present invention will now be described with reference to FIGS. 8-30.

Referring first to FIG. 8, a pictorial representation is shown of a game selection screen 800. The game selection screen 800 is similar to that provided to an LLI subject upon initialization of the computer program according to the present invention. The game selection screen 800 includes the titles of seven computer games that provide distinct training exercises for improving speech recognition in subjects who abnormally process temporal acoustic events, and for building, or rebuilding the neurological connections necessary to accurately process phonemes at the rates common in speech. The game titles include: 1) Old MacDonald's Flying Farm; 2) Block Commander; 3) Circus Sequence; 4) Phonic Match; 5) Phonic Words; 6) Phoneme Identification; and 7) Language Comprehension Builder. Each of these games will be discussed in greater detail below.

When a subject begins execution of the Fast ForWord computer program, he/she is presented with a screen similar to the screen 800. More specifically, upon initiation of the program, the subject is presented with a screen that lists the subjects that are currently being trained by the program. The subject then selects his/her name from the list. Once the subject has selected his/her name, a screen similar to 800 appears, typically listing one of the seven programs, according to a training schedule that is dictated by the program, or is modified by an instructor. The order of the games, and the selection of which one of the seven games that is presented in the screen 800 varies from day to day. The subject then elects to play the first game listed according to the training schedule prescribed for the subject.

In one embodiment, a training schedule is provided by a certified Speech and Language Professional (SLP), and the SLP oversees each training session according to the schedule. An exemplary schedule requires a subject to cycle through five of the seven games for an hour and forty minutes, five days per week, for approximately six weeks. In addition, the schedule typically requires that a subject play Circus Sequence and Language Comprehension Builder everyday, alternating the other games so that they are played approximately the same amount of time.

In an alternative embodiment, the game schedule specified by an SLP at a remote server, and the daily parameters of the schedule are downloaded to the subject's computer, either daily or weekly. The schedule can be optimized over the course of the training program to first develop skills required for subsequent more advanced skills. It can also be used to help manage time in each game so that all of the games are completed at about the same time at the end of the training program. This embodiment allows a subject to obtain the benefits of the Fast ForWord program, and the oversight of a certified SLP, regardless of his/her geographic location. One skilled in the art will appreciate that the training schedule could either be provided in a window on the subject's computer, or could actually control the game selection screen to prompt the user only for those games required on a particular day.

Once a subject selects a particular game, he/she is taken into that particular game's module. Alternatively, once the subject selects his/her name from the list, the particular games may be presented, in a predefined order, without requiring the subject to first select the game. For ease of illustration, each of the seven games will be discussed, in the order represented in FIG. 8.

Referring to FIG. 9, a scene 900 is shown for the first game in the program, Old MacDonald's Flying Farm (OMDFF). OMDFF uses a psychophysical procedure called limited-hold reaction time. A subject is asked to start a trial, in this case by grabbing a flying animal, at which point the game begins presenting a distractor phoneme that is modified in the time domain only. More specifically, information bearing acoustic elements whose temporal location within a phoneme carry important cues for phoneme identification are modified by stretching the acoustic elements in time, say to 150% of their normal duration. The acoustic elements that are stretched include voice onset time (VOT) between consonant and vowel events, as well as fricative-vowel gaps. The inter-stimulus interval (ISI) between presentations of the distractor phoneme is set initially to 500 ms. The distractor phoneme is repeated a random number of times, usually between 3 and 8 times, before the target tone is presented. The target phoneme has normal temporal acoustic parameters. The subject is asked to continue to hold the animal until the target phoneme is presented. When the subject hears the target phoneme, the subject is to release the animal. If the subject accurately hears the target phoneme and releases the animal within a desired "hit" window, then his/her score increases. If the subject misses the target phoneme, the animal flies away and no points are given. As the subject improves, the temporal parameters of the distractor phonemes are reduced in time to that of normal speech, and the ISI is reduced, systematically to 300 ms.

A number of scenes are provided in OMDFF, each correlated to a specific pair of sounds. The correlation of sound pairs to farm scenes is shown below:

So, when a subject grabs the flying animal, the game begins presenting a tone pattern such as: /Si/ . . . /Si/ . . . /Si/ . . . /Si/ . . . /Sti/. When the subject hears /Sti/, the subject is to release the animal.

The scene 900 provides a general farmyard background with three elements that persist across all the scenes. The elements are the score digits 906, the stop sign 908, and the tractor 910. The tractor 910 acts as a progress creature to graphically indicate to a subject their progress during a game. If the subject gets a correct response, the tractor 910 advances across the screen 900, from right to left. The score digits 906 display the subject's current score. The stop sign 908 is common to all seven games, and provides a subject with a means for exiting the game, and then the program.

Also shown on the screen 900 are a flying farm animal 902, and a selection hand 904. In this scene, the flying farm animal 902 is a cow with a rocket pack. Other scenes provide different farm animals propelled through the air with different flying apparatus. Operation of the game OMDFF will now be described with reference to FIG. 10.

In FIG. 10, a flow chart 1000 is provided that illustrates operation of the OMDFF game. The game begins at block 1002 and proceeds to block 1004.

At block 1004, the computer program selects a particular tone sequence to be played for a subject. For example, the program would select the tone pair /Si/ . . . /Sti/, stretched 150%, with an ISI of 500 ms. The tone pair that is selected, the stretching, and the ISI, are all associated with a particular skill level. And, the skill level that is presented to a subject is adapted in real time, based on the subjects ability to recognize the target phoneme, as will be further described below. However, the initial phoneme pair, stretching and ISI are chosen to allow an LLI subject to understand the game, and to begin to distinguish phonemes common in speech. Upon selection of a particular phoneme sequence, and skill level, flow proceeds to block 1006.

At block 1006, the game presents a flying animal 902. As mentioned above, the animal 902 that is presented varies according to which of the phoneme pairs are selected. If the animal 902 is a flying cow, the phoneme pair that will be presented is /Gi/ . . . /Ki/. The animal 902 continues to fly around the screen until the subject places the selection hand 904 over the animal 902, and holds down a selection button, such as a mouse button. After the animal 902 is presented, flow proceeds to decision block 1008.

At decision block 1008, a test is made as to whether the subject has selected the animal 902. If not, flow proceeds to block 1010 where the animal 902 continues to fly. The animal 902 will continue moving about the scene 900 until it is selected. Flow then proceeds to block 1012.

At block 1012, the program begins presenting the selected phoneme sequence. More specifically, an audio formatted file is called by the program that is to be played by a computer, either through speakers connected to the computer, or though headphones worn by a subject. In one embodiment, the file is a QuickTime audio file, configured according to the parameters necessary for the skill level of the user, i.e., phoneme pair, stretching, and ISI. In addition, a starting point in the file is chosen such that the distractor phoneme is presented a random number of times, between 3 and 8 times, before the target phoneme is presented. After the phoneme sequence begins playing, flow proceeds to decision block 1014.

At decision block 1014, a determination is made as to whether the subject has released the animal 902. If the subject has not released the animal 902, a parallel test is made, shown as decision block 1016.

Decision block 1016 tests whether a "hit" window has passed. More specifically, the program contains a lockout window of 200 ms that begins when the target phoneme is played. It is believed that if the subject releases the animal 902 within 200 ms of the target phoneme beginning play, it is merely coincidental that he/she would have heard the target phoneme. This is because no subject's reaction time is quick enough release the animal 902 so soon after hearing the target phoneme. The start of the "hit" window begins after the lockout window, i.e., 200 ms after the target phoneme begins. The end of the hit window is calculated as the start of the hit window, plus the length of one phoneme letter. So, at decision block 1016, if the hit windows has not passed, the computer continues to test whether the subject has released the animal 902. If the hit window has passed, and the subject has not released the animal 902, flow proceeds to block 1026.

At block 1026, a miss is recorded for that test. After recording the miss, flow proceeds back to block 1021.

At block 1021, the skill level for the selected phoneme sequence is decreased, as will be further described below. Flow then proceeds back to block 1006 where another flying animal is presented for the same phoneme sequence.

At decision block 1014, if it is determined that the subject has released the animal 902, instruction flow proceeds to decision block 1018.

At decision block 1018, a determination is made as to whether the hit window has begun. That is, did the subject release the animal 902 during or before the lockout period? If the hit window has not begun, instruction flow proceeds to block 1020.

Block 1020 records a false alarm and instruction flow proceeds to block 1021. It should be appreciated that a false alarm is recorded, rather than a miss, because it suggests that the subject detected a change in the phoneme sequence when a change has not yet occurred. If, at decision block 1018, the hit window has begun, flow proceeds to decision block 1022.

At decision block 1022 a determination is made as to whether the hit window has passed. If the hit window has passed, prior to the subject releasing the animal 902, then flow proceeds to block 1026 where a miss is recorded, as described above. However, if the hit window has not passed flow proceeds to block 1024.

At block 1024, a hit is recorded for the subject. That is, the subject has correctly heard the target phoneme, and has released the animal 902 in an appropriate time frame. Flow then proceeds to decision block 1028.

At decision block 1028, a determination is made as to whether the subject has heard the target phoneme, and released the animal 902 within the hit window, three times in a row. If not, then flow proceeds back to block 1006 where another animal 902 is presented. If the subject has responded correctly, three times in a row, flow proceeds to block 1030.

At block 1030, the skill level for the selected tone sequence is increased by one level. In one embodiment, 18 skill levels are provided for each phoneme sequence. As mentioned above, the skill levels begin temporal modifications of the phonemes, and by separating the presented phonemes with an ISI of 500 ms. As the subject's ability to distinguish between the distractor and target phonemes improves, the temporal modifications of the phoneme is reduced to that of normal speech, and the ISI is reduced to 300 ms. One skilled in the art will appreciate that the degree of phoneme temporal manipulation, from 150% to 100%, the variation of ISI among the skill levels, and the number of skill levels provided, may vary depending on the LLI subject and the type of training that is required. In one embodiment, after a subject successfully passes a phoneme sequence with 150% time modification, and an ISI of 500 ms, the next skill level presented holds the time modification at 150%, but reduces the ISI to 400 ms. Flow then proceeds to decision block 1032.

At decision block 1032 a determination is made as to whether the maximum level has been reached for the selected phoneme sequence. That is, has the subject progressed through all the skill levels to the point that they are correctly recognizing a target phoneme with a duration of 100%, and with an ISI of 0 ms? If not, then flow proceeds to block 1006 where the animal 902 is again presented to the subject, this time, at an increased skill level. However, if the subject has reached the maximum level for a particular phoneme sequence, flow proceeds to block 1004 where a phoneme tone sequence is selected. If a subject has not yet played the new phoneme sequence that is selected, the skill level is set to the easiest level. However, if the subject has previously heard the new phoneme sequence, the level of play begins, either at or below the last skill level obtained, typically 5 skill levels below what was last obtained.

Selection of phoneme sequences and skill levels are performed by the program to insure that a subject is exposed to each of the phoneme pairs, but spends the greater portion of his/her time with those pairs that are the most difficult to distinguish. In addition, the number of recorded hits/misses/false alarms and reaction times are recorded for each level, and for each phoneme pair, on a daily basis. The records are then uploaded to a remote server where they are either reviewed by a remote SLP, or are tabulated and provided to a local SLP. The SLP then has the option of controlling the selection of phoneme sequence selection, and/or skill level, according to the particular needs of the subject, or of allowing automatic selection to occur in a round robin manner.

While not shown, the program also keeps track of the number of correct responses within a sliding window. This is visually provided to a subject by advancing the tractor 910, from the right to the left, for each correct response. After 10 correct responses, creative animations are played, and bonus points are awarded, to reward the subject and to help sustain the subject's interest in the game. Of course, the type of animation presented, and the number of correct responses required to obtain an animation are variables that may be set by an SLP.

Now referring to FIG. 11 a screen 1100 is shown of the second game in the Fast ForWord program, entitled Block Commander. The Block Commander game presents a subject with audio prompts, directing the subject to perform an action. An exemplary action might be "point to the green circle." The types of prompts are grouped according to difficulty, requiring a subject to perform increasingly sophisticated tasks, depending on their skill level. If the subject responds correctly he/she is awarded a point. Otherwise, the cursor hand turns red and demonstrates how the command should have been performed. This feedback allows the subject to learn from the computer the more difficult manipulations that are required. In addition, the prompts are digitally processed by stretching the speech commands (in the time domain), and by emphasizing particular frequency envelopes in the speech, that contain time modulated acoustic components.

The screen 1100 contains a number score 1102 and a stop sign 1104. The number score 1102 provides visual feedback to a subject regarding their progress in the game, and the stop sign 1104 provides a selection mechanism for ending the game. Also shown is a cat 1106. The cat 1106 provides animations for a subject during training. A grid 1120 is shown, in a 55 degree perspective, upon which are placed 3D tokens, further described below. In the center of the grid 1120 is an ear/hand button 1108. When a subject places a hand selector 1110 on top of the ear/hand button 1108, and selects the icon (by pressing a mouse key), then a trial in the Block Commander game begins. This is shown in FIG. 12, to which attention is now directed.

In FIG. 12, a screen shot 1200 is shown that includes the stop sign, number score, and grid, as shown above. In addition, a row of different colored squares 1202, and a row of different colored circles 1204 are provided. Use of the squares 1202 and the circles 1204 will be described below with reference to FIG. 13. Also shown are a number of progress tokens 1206 at the bottom of the screen 1200. The progress tokens 1206 indicate the number of correct answers within a particular instance of the game. In one embodiment, after 5 tokens 1206 are shown, indicating 5 correct responses, a reward animation and bonus points are provided to the user.

Now referring to FIG. 13, a flow chart 1300 is shown that illustrates operation of the Block Commander game. Execution begins at block 1302 and proceeds to block 1304.

At block 1304 the game selects the first playing level that is to be presented to a subject. To the right of block 1304 is a table 1330 that illustrates the 5 processing levels that are used in the Block Commander game. The levels are distinct from each other in terms of the amount of stretching (in the time domain) that is used on speech, and the amount of emphasis that is applied to selected frequency envelopes within the speech. Flow then proceeds to block 1306.

At block 1306, the game presents a program to a subject that trains the subject to play the game. The training portion consists of 3 rounds. The first round trains the subject to distinguish between object sizes, e.g., large and small. The second round trains the subject to distinguish between object shapes, e.g., square and circle. The third round trains the subject to distinguish between object colors, e.g., blue, red, yellow, green and white. More specifically, the prompts given to a subject during training are:

For a subject to pass any of the training rounds, and progress to the next training round, two correct hits are required for each command prompt, with no errors. If an error is made, the score is reset, and play for that round starts over. All of the prompts for the training rounds are at processing level 1, 150% duration and 20 dB emphasis. After a subject has completed the training program he/she will not see it again. Upon completion of the training program, flow proceeds to decision block 1308.

At decision block 1308 a determination is made as to whether the training has been completed. If not, then flow proceeds back to block 1306 where training continues. If training has been completed, flow proceeds to block 1310.

At block 1310, a warm up exercise is presented to a subject. The warm up exercise is presented each time a user plays the game, at the speech processing level that was last completed. The warm up round includes the following prompts:

The ordering of the prompts is random each time the warm up is played. After presentation of each of the prompts flow proceeds to decision block 1312.

At decision block 1312, a determination is made as to whether the warm up round has been completed. If not, then flow proceeds back to block 1310 where the warm up continues. Otherwise, flow proceeds to block 1314.

At block 1314, an appropriate processing level is selected for a subject. The first time a subject plays the Block Commander game, processing level 1 is selected. However, after the subject has progressed beyond processing level 1, the level selected will be the level that the subject last played. Flow then proceeds to block 1316.

At block 1316, the first round of the game is presented to a subject. As mentioned above, in one embodiment of the Block Commander game, six rounds are provided. The rounds are as follows:

Each of the prompts are presented to the user in random order, but successful completion of each of the prompts in a round is required before a round is considered complete. After a first prompt is provided to a subject, flow proceeds decision block 1318.

At decision block 1318, a determination is made as to whether there have been 90% correct responses in a sliding group of 5 items. If not, then flow proceeds back to block 1316 where another prompt in a round is provided. If there have been 90% correct responses, as will be illustrated by 5 progress tokens at the bottom of the screen, then flow proceeds to block 1320.

At block 1320, the subject is shown a reward animation. In one embodiment, the animation consists of characters morphing out of the blocks on the board. Flow then proceeds to decision block 1322.

At decision block 1322, a determination is made as to whether the round is complete. A round is complete when a subject successfully responds to all of the prompts in the round. If the round is not complete, flow proceeds back to block 1316 where another prompt is provided to the subject. If the round is complete, flow proceeds to decision block 1324.

At decision block 1324, a determination is made as to whether all six rounds within the game have been completed. If not, then flow proceeds to block 1326 where the round level is incremented. Flow then proceeds back to block 1316 where prompts for the new round are presented. If decision block 1324 determines that all rounds have been completed, flow proceeds back to block 1314 where an appropriate skill level is selected. In one embodiment, if a subject successfully completes all six rounds, at skill level 1 (150% duration, 20 dB emphasis), he/she will progress to skill level 2 (125% duration, 20 dB emphasis).

The Block Commander program begins by providing a subject with a number of simple commands, stretched in time, with particular emphasis given to phoneme components that are difficult for an LLI subject to understand. As the subject correctly responds to the simple commands, the commands increase in difficulty. Once the subject masters the more difficult commands, the amount of stretching, and the amount of emphasis is reduced, and the process is repeated. The rounds continue, over the course of days and weeks, until the subject is correctly responding to the difficult commands at skill level 5, which is normal speech.

One skilled in the art will appreciate that the commands cause the subject, not only to understand the phonemes that are presented, but also to apply logical reasoning to the more difficult commands, and to recall the constructs of the commands. The requirement that the subject recall the command constructs is directed at improving the subjects memory, as well as to improving their ability to process acoustic events. It is believed that the games repetitive nature, that trains the subject's neurological connections to process speech, is also helpful in improving the subject's memory, and his/her cognitive skills in understanding linguistic relationships.

Now referring to FIG. 14, a screen shot 1400 is shown for the third game in the Fast ForWord program, entitled Circus Sequence. The Circus Sequence game trains a subject to distinguish between upward and downward frequency sweeps that are common in the stop consonant portion of phonemes, by varying the duration and frequency of the sweeps, and by varying the inter-stimulus interval (ISI) between presentation of the sweeps.

The screen 1400 contains a number score 1402, a stop sign 1404, and a progress element 1406, all within a circus ring environment. In addition, the screen 1400 contains a hand selector 1408, and an ear/hand button 1410. As in the Block Commander game, a user begins a test by selecting the ear/hand button 1410 with the hand selector 1408.

Referring to FIG. 15, a screen shot 1500 is shown that illustrates two elements 1502, 1504 that are presented to a subject after the ear/hand button 1410 is selected. The left element 1502 pertains to an upward frequency sweep, and the right element 1504 pertains to a downward frequency sweep. In addition, a progress element 1506 is shown elevated above the circus ring floor, to indicate that a subject has correctly responded to a number of tests. Game play will now be illustrated with reference to FIG. 16.

FIG. 16 provides a flow chart 1600 that illustrates program flow through the training portion of the Circus Sequence Game. Training begins at block 1602 and proceeds to block 1604.

At block 1604, the program begins presenting a random sequence of frequency sweeps to a subject. All sweep sequences are of the form: up-up; up-down; down-up; or down-down. Thus, if the program presents the sweep sequence "up-up", a subject is to click on the left element 1502 twice. If the program presents a sweep sequence "down-up", the subject is to click on the right element 1504, then on the left element 1502. So, once the program provides a sweep sequence to the subject, the subject selects the elements corresponding to the frequency modulated (FM) tone sequence. If the subject is correct, he/she is awarded points, the progress element 1506 advances upwards, and the ear/hand button 1410 is presented, allowing the subject to begin another test. During training, all upward sweeps are presented starting at 1 kHz and all downward sweeps ending at 1 kHz, with upward/downward sweeps at 16 octaves per second. The duration of the sweeps are 80 ms, and the sweeps are separated by 1000 ms. Research has shown that most LLI subjects are capable of distinguishing between frequency sweeps of this duration, and having an ISI of 1000 ms. After each sweep sequence is presented, flow proceeds to decision block 1606.

At decision block 1606, a determination is made as to whether the subject has correctly responded to 80% of the trials over a sliding scale of the last ten trials. If not, then flow proceeds back to block 1604 where the sequences continue to be presented. If the subject has correctly responded 80% of the time, flow proceeds to block 1608.

At block 1608, random sequences are again presented, at 1 khz, having a duration of 80 ms and an ISI of 1000 ms. Flow then proceeds to decision block 1610.

At decision block 1610, a determination is made as to whether the subject has correctly responded to 90% of the trials over a sliding scale of the last ten trials. If not, then flow proceeds to decision block 1612. If the subject has correctly responded to 90% of the trials over a sliding scale of the last ten trials, flow proceeds to block 1614.

At decision block 1612, a determination is made as to whether a subject has correctly responded to less than 70% of the trials, over a sliding scale of the last 20 trials. If not, indicating that he/she is responding correctly between 70-90% of the time, then flow proceeds back to block 1608 where the sweep sequences continue to be presented. If a determination is made that the subject is correctly responding less than 70% of the time over the last 20 trials, then flow proceeds back to block 1604, where the training begins again.

At block 1614, a 3-up, 1-down rule begins. This rule allows a subject to advance in difficulty level every time 3 correct responses are provided, while reducing the level of difficulty any time an incorrect response is given. Research has shown that a 3-up, 1-down rule allows a subject to obtain a correct response rate of approximately 80% near threshold, which is desired to motivate and encourage the subject to continue. A reduced accuracy rate discourages a subject, a situation that is not desired especially if the subject is an LLI child. Once the 3-up, 1-down rule is started, flow proceeds to decision block 1616.

At decision block 1616, a determination is made as to whether a subject has responded correctly the last 3 tests. If so, then flow proceeds to block 1620. If not, then flow proceeds to decision block 1618.

At decision block 1618, a determination is made as to whether a subject has incorrectly responded to the last test. If not, then flow proceeds back to decision block 1616 where another test is provided. However, if the subject has incorrectly responded to the last test, the difficulty level is reduced one level, and flow proceeds back to decision block 1616 where another test is presented. During the training level, all tests are performed at 80 ms duration, with 1000 ms ISI, which is the easiest skill level. Therefore, if the subject incorrectly responds at that level, no change in difficulty is made.

At block 1620, the skill level is increased. During training, the sweep sequences are presented at 1 khz, with 80 ms duration, but the ISI is reduced between the sweeps each time the level is incremented. In one embodiment, the ISI levels start at 1000 ms, and proceed through 900 ms, 800 ms, 700 ms, 600 ms and 500 ms. Flow then proceeds to decision block 1624.

At decision block 1624, a determination is made as to whether the ISI is at 500 ms. If not, then flow proceeds back to decision block 1616 where sweep sequences continue to be presented. If the ISI is 500 ms, the training session ends and the subject is allowed to enter the real game, at block 1626.

Referring now to FIG. 17, a flow chart 1700 is provided that illustrates operation of the Circus Sequence game, after the training session has been completed. The game begins at block 1702 and proceeds to block 1704.

At block 1704, an appropriate skill level is selected. The skill levels used by Circus Sequence are shown in table 1730. For each of three frequencies: 500 hz, 1 khz, and 2 khz, a number of skill levels are provided. The skill levels begin by presenting frequency sweeps having a duration of 80 ms, and an ISI between the sweeps of 500 ms. As a subject advances, the ISI is reduced, either to 0 ms, or in one embodiment, to 125 ms. It should be appreciated that the ISI increments used should be selected to slowly train a subject's ability to distinguish between similar phonemes, such as /ba/ and /da/, while not frustrating the subject by training beyond levels required to distinguish between such phonemes.

When a subject first plays Circus Sequence, after passing training, he/she is provided with frequency sweeps beginning at 1 khz, having 80 ms duration and an ISI of 500 ms. On subsequent days, the frequency that is selected is random, and can be either 500 hz, 1 khz or 2 khz. Once the appropriate skill level has been selected, flow proceeds to block 1706.

At block 1706, a tone sequence is presented, according to the selected skill level. Flow then proceeds to decision block 1708.

At decision block 1708, a determination is made as to whether the subject has correctly responded to the last 3 trials. If not, then flow proceeds to decision block 1710. If the subject has correctly responded to the last 3 trials, flow proceeds to block 1712.

At decision block 1710, a determination is made as to whether the subject has incorrectly responded to the last trial. If not, then flow proceeds back to block 1706 where another tone sequence is presented. If the subject incorrectly responded to the last trial, flow proceeds to block 1714.

At block 1714, the skill level is decremented. If the skill level has an ISI of 500 ms, no decrease is made. However, if the skill level has an ISI that is less than 500 ms, the difficulty is reduced 1 level. For example, if the subject incorrectly responds to a trial having an ISI of 180 ms, for example, the difficulty level will be reduced, so that the next tone sequence will have an ISI of 185 ms. Flow then proceeds back to block 1706 where another tone sequence is presented.

At block 1712, if the user has correctly responded to the last 3 trials, the skill level is incremented. For example, if a subject is at a skill level with a sweep duration of 80 ms and an ISI of 250 ms, the skill level will increase such that the ISI for the next tone sequence will be 200 ms. Flow then proceeds to decision block 1716.

At decision block 1716, a determination is made as to whether the ISI is at 150 ms. If not, then flow proceeds to decision block 1720. If the ISI is at 150 ms, flow proceeds to block 1718.

At block 1718, the next lower duration is enabled. This allows the program to simultaneously trial a subject with multiple sweep durations, once the subject is successfully responding at an ISI level of 150 ms. For example, if a subject is correctly responding to tone sequences of duration 80 ms, with an ISI of 150 ms, then testing continues at 80 ms. In addition, testing is begun with sweep sequences of duration 60 ms, at an ISI of 500 ms. Flow then proceeds to back to block 1706 where another tone sequence is presented. This allows the program to present tone sequences of different duration, and different ISI, while tracking progress for each duration/ISI combination.

At decision block 1720, a determination is made as to whether the subject has reached a training threshold. In one embodiment, a training threshold is reached when the subject has had eight skill level reversals within six skill levels of each other. If such a threshold is reached, flow proceeds to block 1721. Otherwise, flow proceeds to decision block 1722.

At block 1721, the program moves the subject to the next frequency category to be tested. It is believed that once a threshold has been met on a particular day, the subject should not continue being tested at the same frequency. Thus, the program allows a subject to progress, either to an ISI of 0 ms (or some other minimal ISI) or to a threshold at one frequency, and then begin testing at an alternative frequency. Flow then proceeds back to block 1706.

At decision block 1722, a determination is made as to whether the ISI for a particular tone duration is 0 ms. If not, then flow proceeds back to block 1706 where another sweep sequence is presented. However, if a subject has reached a skill level of 0 ms ISI for a particular duration, flow proceeds to block 1724.

At block 1724, the program deletes the duration associated with the 0 ms ISI from the trial. This is because testing at that level is no longer required by the subject due to their proficiency. However, as mentioned above, an alternative embodiment may select an ISI of greater than 0 ms as the point where the duration is deleted from the game. Flow then proceeds back to block 1706 where more tone sequences are presented.

While not shown, in one embodiment, a threshold level is provided that causes the game to begin testing a subject at an alternate frequency. For example, if the subject is testing at 500 hz, and a threshold is reached, the program will begin testing the subject at 2 khz. The threshold is reached when a subject has 8 skill level reversals within 6 levels of each other. When this occurs, the program ceases testing at the frequency for which the threshold was reached, and begins testing at an alternative frequency.

Also, when a subject begins each day of testing, a frequency different than that tested the previous day is begun. Moreover, a skill level that is 5 less than completed the previous day is chosen, presuming the subject completed at least 20 trials for that frequency.

As mentioned above, each correct response causes the progress element 1506 to advance upward. After ten correct responses, a reward animation is provided to entertain the subject. When the animation ends, the subject is prompted with the ear/hand button 1410 to begin another trial.

Now referring to FIG. 18, a screen shot 1800 of the fourth game in Fast Forward, Phonic Match, is provided. The screen 1800 includes a set of pictures 1802, a progress creature 1804, a stop sign 1806, and a number score 1808. The progress creature 1804, stop sign 1806 and number score 1808 function similarly to those described in previous games.

The set of pictures 1802 are arranged into a 2.times.2 grid. When a subject selects any of the pictures, a word or phoneme is played. On any grid, there are two pictures that play the same word. Thus, for a 2.times.2 grid, there are two words that will be presented. The test for the subject is to distinguish between similar words, to recall which picture is associated with which word, and to sequentially select two pictures that present the same word. Similar words are presented together, with the words processed according to the processing levels shown in table 1902 of FIG. 19.

Initially, subjects are presented words at processing level 1, with a duration of 150%, and having 20 dB emphasis of selected frequency envelopes within the words. In addition, different skill levels, as shown in table 1904, are provided that increase the grid size for a particular trial, and set the maximum number of clicks, or selections, that a subject can attempt before losing the trial. Operation of the game is illustrated in FIG. 20. However, before providing a detailed description of game operation, the words used in the game are shown.

Referring now to FIG. 20, the Phonic Match game begins at block 2002, and proceeds to block 2004.

At block 2004, a 2.times.2 grid is presented. The words associated with the 2.times.2 grid are selected from one of the four Word Groups shown above. The selection of the Word Group is random, except that tracking of previously played Word Groups is done to insure that all Word Groups are equally represented, and that a subject is not provided the same Word Group as played on an immediately preceding day. The words within a Word Group are typically selected according to their acoustic similarity.

The subject is required to sequentially select two pictures that have the same word associated with them. When a subject sequentially selects two pictures associated with the same word, the pictures are removed from the gird being played. After a subject completes a 2.times.2 grid, whether correctly or incorrectly, flow proceeds to decision block 2006.

At decision block 2006, a determination is made as to whether the subject has successfully passed three 2.times.2 grids. Referring to table 1904 of FIG. 19, ten skill levels are shown. When a 2.times.2 grid is first presented, the skill level entered is level 8. Skill level 8 defines a 2.times.2 grid, with a maximum number of allowed clicks as 8. If a subject selects pictures on a 2.times.2 grid more than 8 times, the grid is not considered passed, and game flow proceeds back to block 2004 where another grid is presented. If not, then flow proceeds back to block 2004 where another 2.times.2 grid is presented with words from the same Word Group. If the subject has successfully passed three 2.times.2 grids, thus progressing from level 8 through level 10, flow proceeds to block 2008.

At block 2008, a new grid is presented for a particular Word Group, or stimulus set. Initially, a 3.times.3 grid is provided, at skill level 2. The maximum number of clicks allowed for a subject to pass a 3.times.3 grid is 20. Within a 3.times.3 grid, 1 of the pictures is a wildcard, since there are an odd number of pictures. Selection of the wildcard simply removes the picture from the grid, and does not count against the subject as a selection, or click. After a 3.times.3 grid is presented to a subject, flow proceeds to decision block 2010.

At decision block 2010, a determination is made as to whether the subject passed the level. That is, did the subject properly distinguish between word pairs, and sequentially select picture pairs associated with words in 20 or less clicks. If so, then flow proceeds to block 2012. If not, then flow proceeds to block 2014.

At block 2012, the skill level is incremented. For example, if a subject was at level 2, he/she will increment to level 3. Note: levels 2-3 present a 3.times.3 grid with a maximum number of clicks of 20, while levels 4-7 present a 4.times.4 grid with a maximum number of clicks of 60. Once the skill level is incremented, flow proceeds to block 2020.

At block 2020, a grid according to the new skill level is presented. The grid is associated with the same Word Group that was previously used, but possibly with different words from the group. Flow then proceeds to decision block 2022.

At decision block 2022, a determination is made as to whether the subject has passed the level. That is, did the subject correctly associate the word pairs in less than or equal to the number of allowed clicks. If not, flow proceeds to block 2014. If the subject passed the level, flow proceeds to decision block 2024.

At decision block 2024, a determination is made as to whether the subject has reached skill level 7. Level 7 is termed the "decision" level. If the skill level that has just been passed is not level 7, then flow proceeds back to block 2012 where the skill level is incremented. However, if the skill level passed is level 7, flow proceeds to decision block 2026.

At decision block 2026, a determination is made as to whether all four stimulus sets, or Word Groups have been passed. If not, then flow proceeds to block 2018. However, if a subject has correctly passed skill level 7, for all four Word Groups, flow proceeds to block 2028.

At block 2028, the next processing level is selected. Referring to table 1902 of FIG. 19, a subject begins at processing level 1 (duration 150%, emphasis 20 dB). Once all four Word Groups have been passed at skill level 7, the amount of audio processing to the words is reduced. First, the duration of the words is reduced, from 150%, to 125%, to 100%, and then the amount of emphasis applied to selected frequency components is reduced, from 20 dB, to 10 dB, to 0 dB. Once a subject has reached processing level 5, he/she is presented with normal speech. After the next processing level is selected, flow proceeds to decision block 2030.

At decision block 2030, a determination is made as to whether all processing levels have been completed. That is, has the subject reached processing level 5. If not, flow proceeds back to block 2004 where the game begins anew, with a 2.times.2 grid, but at the new processing level. However, if the subject has reached processing level 5, flow proceeds to block 2032.

At block 2032, a 5.times.5 grid is provided, with a maximum number of allowable clicks as 90. From this point forward, the game continues playing indefinitely, but the decision round, level 7, switches from a 4.times.4 grid to a 5.times.5 grid.

Referring back to decision block 2022, if a subject does not pass a particular level, flow proceeds to block 2014.

At block 2014, the skill level is decremented. Flow then proceeds to decision block 2016.

At decision block 2016, a determination is made as to whether the new skill level is less than level 1. Level 1 is considered a "slip" level indicating that if a user failed at this level, a new Word Group should be provided. If the skill level is not less than 1, flow proceeds back to block 2020 where a new grid is presented, according to the present level. If the new level is less than 1, that is, if the subject failed to pass a grid, at skill level 1, flow proceeds to block 2018.

At block 2018, the program discontinues presenting words from the present Word Group, and changes the Word Group used for the grids. Flow then proceeds back to block 2008 where a 3.times.3 grid is presented, at skill level 2, using words from the new Word Group.

The flow chart 2000 demonstrates that a subject is required to proceed from level 2 through level 7 for each of the four Word Groups, at a particular processing level, before he/she is allowed to advance to the next processing level. The progress creature descends with each click. If the creature reaches the bottom, then the grid is not passed. If all picture pairs are matched prior to the creature reaching the bottom, extra points are awarded, a reward animation is presented and the grid is considered passed. When a subject has correctly selected a predetermined number of picture pairs, the progress animal 1804 reaches the top, and the subject is rewarded by an animation.

Referring now to FIG. 21, a screen shot 2100 is shown illustrating the fifth game in the Fast Forward program, entitled Phonic Words. Phonic Words presents a subject with a sentence prompt that requires the subject to distinguish between two similar words, to accurately select one of two pictures 2108, 2110, using a selection hand 2112. The table below provides a list of the word pairs used. The first word in the pair is always the correct answer, but its representational image could appear on the left or right of the screen 2100.

base-face, face-base, vase-base, base-vase, face-vase, vase-face, bee-me, me-bee, knee-bee, bee-knee, knee-me, me-knee, breathe-breeze, breeze-breathe, day-they, they-day, lawn-yawn, yawn-lawn, ache-lake, lake-ache, ache-rake, rake-ache, ache-wake, wake-ache, lake-rake, rake-lake, lake-wake, wake-lake, rake-wake, wake-rake, sink-think, think-sink, chip-dip, dip-chip, sip-zip, zip-sip, chip-sip, sip-chip, chip-zip, zip-chip, dip-sip, sip-dip, dip-zip, zip-dip, pack-shack, shack-pack, tack-shack, shack-tack, pack-tack, tack-pack, tack-tag, tag-tack, rung-young, young-rung, rung-run, run-rung, young-run, run-young, pat-path, path-pat, bear-bell, bell-bear, thumb-tongue, tongue-thumb, comb-cone, cone-comb, mouse-mouth, mouth-mouse, cash-catch, catch-cash, fan-fang, fang-fan, sauce-saws, saws-sauce, bass-bath, bath-bass, cheese-chief, chief-cheese, foam-phone, phone-foam, fuzz-fudge, fudge-fuzz, safe-shave, shave-safe, long-lawn, lawn-long, piece-peas, peas-piece, piece-peach, peach-piece, peas-peach, peach-peas, wash-watch, watch-wash.

As before, the screen 2100 contains an ear/hand button 2102 for beginning a trial, a stop sign 2104 for ending the game, and a number score 2106. Within the number score 2106 are five acorns, indicating the processing level currently being tested. Also shown are progress creatures 2114 indicating a number of correct responses. As a subject correctly responds to the game, a new progress creature 2114 is added. When the number of progress creatures 2114 reaches ten, a reward animation is provided and bonus points are awarded.

Referring to FIG. 22, a screen shot 2200 is shown where the word pair peach-peas is being tested. After a subject listens to a prompt containing the target word, he/she selects one of the two pictures. The subject, whether correct or incorrect, will then be shown the correct selection, in this case peach, by having the mask removed from the picture frame 2202.

Referring now to FIG. 23, operation of the Phonic Words game is illustrated by flowchart 2300. Please note that five processing levels, similar to those used above in Phonic Match and Block Commander, are shown in table 2340. The game begins at block 2302 and proceeds to training block 2304.

At training block 2304 the subject is prompted to "press the ear button". The prompting is processed at level 1 (duration 150%, emphasis 20 dB). Flow then proceeds to decision block 2306.

At decision block 2306, a determination is made as to whether the ear/hand button 2102 has been pressed. If not, then flow proceeds back to block 2304 where the prompting is repeated. If the ear/hand button 2102 has been pressed, flow proceeds to block 2308.

At block 2308, praise is played for the subject. Flow then proceeds to block 2310.

At block 2310, a single image appears in one of the two frames 2108, 2110, and a sound file pertaining to the image is played for the subject. Flow then proceeds to decision block 2312.

At decision block 2312, a determination is made as to whether the subject has selected the appropriate image. The image continues to be displayed until the subject selects the image. Flow then proceeds to decision block 2314.

At decision block 2314, a determination is made as to whether the subject has correctly selected the single image, three times. If not, then flow proceeds back to block 2310 where another image is presented, with its associated word. If the subject correctly selects an image/word combination three times, flow proceeds to block 2316.

At block 2316, a pair of images are presented, along with a command prompt containing a word associated with one of the images. The other image presented is termed the distractor image. The user must click on the correct image 4 out of 5 times in a sliding scale to start the game. After the double image is presented, flow proceeds to decision block 2318.

At decision block 2318, a determination is made as to whether the subject has correctly selected an image, from the image pair, in 4 out of 5 cases, on a sliding scale. If not, then flow proceeds back to block 2316 where another image pair is presented. Otherwise, flow proceeds to block 2320 where the subject enters the game. Flow then proceeds to block 2322.

At block 2322, a subject is presented a sequence of image pairs, with associated words selected from a particular processing set. The processing sets are chosen by grouping words having similar phoneme characteristics. Once all of the words have been presented within a processing set, flow proceeds to decision block 2324.

At decision block 2324, a determination is made as to whether the subject has correctly understood a word, and properly selected its associated picture from the picture pair with 90% or greater accuracy. If not, flow proceeds back to block 2322 where random selection of image/word pairs continue, until a 90% success rate is achieved. Flow then proceeds to block 2326.

At block 2326, a new processing set is selected. Flow then proceeds to decision block 2328.

At decision block 2328, a determination is made as to whether all of the processing sets have been completed. If not, then flow proceeds back to block 2322 where random selection of image/word pairs are presented from the current processing set. However, if all of the processing sets have been completed, flow proceeds to block 2330.

At block 2330, the processing level is incremented. Initially, the processing level is level 1. After a subject has completed all of the processing sets, with a 90% or greater accuracy for each of the sets, the processing level is increased to level 2. As described above, the duration of the words is decreased first, from 150%, to 125% to 100%, and then the emphasis of selected frequency envelopes is reduced, from 20 dB, to 10 dB, to 0 dB, until normal speech (level 5) is obtained. After the processing level is incremented, flow proceeds to decision block 2332.

At decision block 2332, a determination is made as to whether a subject has completed all of the sets at processing level 5. If not, then flow proceeds back to block 2322 where random selection of image/word pairs within a set are presented at the new processing level. However, if the subject has completed all of the processing sets at level 5, flow proceeds to block 2334.

At block 2334, Phonic Words continues to drill the subject randomly selecting image/word pairs within a processing set, at level 5.

Now referring to FIG. 24, a screen shot 2400 is provided for the sixth game in the Fast ForWord program, entitled Phoneme Identification. Phoneme Identification processes a number of phoneme pairs by selectively manipulating parameters such as consonant duration, consonant emphasis, and inter-stimulus interval. More specifically, five phoneme pairs are tested, each pair containing a target sound and a distractor. These include: 1) aba-ada; 2) ba-da; 3) be-de; 4) bi-di; and 5) va-fa.

For each phoneme pair, 26 different skill levels are provided, each level differing from the other in the degree of processing applied (duration and emphasis), and in the separation (ISI) of the distractor and target phoneme. Skill level 1 processes the phoneme pair by stretching the consonant portion 150% while leaving the vowel portion untouched, emphasizing selected frequency envelopes in the consonant portion 20 dB, and separating the distractor and target phonemes by 500 ms, for example. Skill level 26 provides a phoneme pair without stretching or emphasis, and with an ISI of 0 ms. Skill levels 2-25 progress towards normal speech by applying less and less consonant processing, with less and less separation between the distractor and target phonemes.

The screen 2400 contains an ear/hand button 2402 to allow a subject to begin a trial, a number score 2404 for tracking correct responses, a stop sign 2406 for exiting the game, a hand selector 2408, and progress elements 2410 for graphically illustrating progress to a subject. When the game is initially selected, five different animals are shown on the screen, each pertaining to a phoneme pair to be tested. A subject may select any one of the five animals to begin the game. After a subject has played the game with one of the five animals, the choice is reduced to four animals, and so on.

Referring to FIG. 25, a screen shot 2500 is shown with two polar bears 2502, 2504. In one embodiment, the polar bears 2502, 2504 are associated with the phoneme pair ba-da. There are five background scenes, each associated with an animal/phoneme pair, each having their own animations, etc. When a subject presses the ear/hand button 2402, the game plays a target phoneme, either ba or da. The phoneme pair is then presented by the polar bears 2502, 2504 with one bear speaking the distractor and the other bear speaking the target sound. A subject is required to distinguish between the distractor and target phonemes, and to select with the hand selector 2508, the polar bear that spoke the target phoneme. Details of how the game Phoneme Identification is played will now be provided with reference to FIGS. 26 and 27.

Referring to FIG. 26, a flow chart 2600 is shown that illustrates the training module of the Phoneme Identification game. Training begins at block 2602 and proceeds to block 2604.

At block 2604, the game presents the screen shot 2400, and prompts a subject to "press the ear button". Flow then proceeds to decision block 2606.

At decision block 2606, a determination is made as to whether the subject has pressed the ear/hand button 2402. If not, then flow proceeds back to block 2604 where the prompt is repeated, after a predetermined interval. If the subject has pressed the ear/hand button 2402, flow proceeds to block 2608.

At block 2608, the ear/hand button 2402 is presented, but this time without an audio prompt. Flow then proceeds to decision block 2610.

At decision block 2610, a determination is made as to whether the subject has pressed the ear/hand button 2402. If not, then flow proceeds back to block 2608. The subject remains in this loop until the ear/hand button 2402 is pressed. Once the ear/hand button 2402 is pressed, flow proceeds to block 2612.

At block 2612, a target phoneme, pertaining to a selected animal pair, is played for a subject. The target phoneme is processed at level 1, 150% duration, with 20 dB emphasis, as shown by the table 2640. Flow then proceeds to block 2614.

At block 2614, a single animal is presented that speaks the target phoneme. Flow then proceeds to decision block 2616.

At decision block 2616, a determination is made as to whether the animal that spoke the target phoneme has been selected. If not, flow proceeds back to block 2614 where the animal again speaks the target phoneme, after a predetermined interval. However, if the subject has selected the animal, flow proceeds to decision block 2618.

At decision block 2618, a determination is made as to whether the subject has correctly pressed the animal in ten trials. If not, then flow proceeds back to block 2612 where another trial is begun. However, once the subject has correctly responded in ten trials, flow proceeds to block 2620.

At block 2620, a target phoneme is again presented, at level 1 processing. Flow then proceeds to block 2622.

At block 2622, two animals are now presented, one speaking the target phoneme, the other speaking the distractor phoneme. The order of speaking the target and distractor phonemes is random, with the animal on the left speaking first, and the animal on the right speaking last. However, in this training level, the animal that speaks the target phoneme is visually highlighted for the subject. Both the target and distractor phonemes are processed at level 1, and are separated in time by 500 ms. Flow then proceeds to decision block 2624.

At decision block 2624, a determination is made as to whether the subject has correctly selected the animal speaking the target phoneme in 8 out of 10 trials, on a sliding scale. If not, then flow proceeds back to block 2620 where another trial is begun. If the subject has correctly responded in 8 out of 10 trials, flow proceeds to block 2626.

At block 2626, a target phoneme is presented to a subject, processed at level 1. Flow then proceeds to block 2628.

At block 2628, two animals are shown presenting a target phoneme and a distractor phoneme, both processed at level 1, with an ISI of 500 ms. The order of target/distractor phonemes is random. For this trial, however, the animal speaking the target phoneme is not visually highlighted for the subject. Flow then proceeds to decision block 2630.

At decision block 2630, a determination is made as to whether the subject has correctly responded to 8 out of 10 trials, on a sliding scale. If so, then the subject has successfully completed the training and flow proceeds to block 2634, allowing the subject to advance to the game. However, if the subject has not been successful in 8 out of 10 trials, then flow proceeds to decision block 2632.

At decision block 2632, a determination is made as to whether the subject has responded correctly less than 70% of the time in at least 10 trials. If not, then flow proceeds back to block 2626 where another trial is presented. If the subject has less than a 70% success rate, over at least 10 trials, then flow proceeds back to block 2614 where trials begin again, but where visual highlighting of the animal speaking the target phoneme is provided for the subject.

Referring now to FIG. 27, a flow chart 2700 is provided that illustrates play of the Phoneme Identification game. Play begins at block 2702 and proceeds to decision block 2704.

At decision block 2704, a determination is made as to whether the ear/hand button 2402 has been pressed. If not, then flow proceeds back to decision block 2704 until the subject chooses to hear the target phoneme. If the ear/hand button 2402 has been pressed, flow proceeds to block 2706.

At block 2706 a target phoneme is presented at an appropriate processing level. If this is the first time a subject has played the game, then the processing level for the phonemes is level 1, and the ISI between the target and distractor phonemes is 500 ms. Otherwise, the skill level pertains to the historical success of the subject, with the particular phoneme pair, as will be further described below. Flow then proceeds to block 2708.

At block 2708, two animals are shown, corresponding to the phoneme pair being tested, speaking the processed target and distractor phonemes, in random order. Flow then proceeds to decision block 2710.

At decision block 2710, a determination is made as to whether the subject has correctly selected the animal speaking the target phoneme. If not, then flow proceeds to block 2720. If the subject has correctly responded to the trial, flow proceeds to decision block 2712.

At block 2720, the skill level for play is decremented. For example, if the processing level is at level 1, having consonant duration of 150%, and emphasis of 20 db, but the ISI between the target and distractor phonemes is at 100 ms, the game will drop back to a skill level where the ISI is at 110 ms. However, if the skill level of play is already at level 1, then no change in processing is made.

At decision block 2712, a determination is made as to whether the subject has correctly responded in the last 3 consecutive trials. If not, then flow proceeds back to decision block 2704, awaiting another trial to begin. However, if the subject has correctly responded to the last 3 trials, flow proceeds to block 2714. It should be appreciated that the procedure illustrated in blocks 2710-2712 is the 3-up, 1-down rule, previously described in the Circus Sequence game above.

At block 2714, the skill level of the game is incremented. For example, if a subject has correctly responded to 3 consecutive trials, and is at a processing level of 100% duration, 20 dB emphasis, and an ISI of 0 ms, the next level of play will be at 100% duration, 10 dB emphasis, and an ISI of 500 ms. Flow then proceeds to decision block 2716.

At decision block 2716, a determination is made as to whether the highest skill level has been reached. If the subject has correctly responded to the last 3 trials, with no processing of the phonemes, and with minimal ISI between the target and distractor, then flow proceeds to block 2718. Otherwise flow proceeds to decision block 2722.

At decision block 2722, a determination is made as to whether the subject has reached a threshold. In one embodiment, a threshold is reached if the subject has had 8 skill level reversals within 6 skill levels of each other. If the subject has not reached a threshold, flow proceeds back to block 2704 where another trial is begun. If the subject has reached a threshold, flow proceeds to block 2718.

At block 2718, a new stimulus category is selected. That is, a new phoneme pair is selected for testing. Thus, if the subject has been tested with the phoneme pair ba-da, and has either mastered the pair by reaching the highest skill level, or has reached a threshold, then an alternate phoneme pair is selected, say aba-ada. Flow then proceeds back to block 2704 where a trial awaits using the new phoneme pair. In one embodiment, the skill level used for the new phoneme pair is selected to be 5 less than previously achieved for that pair. Or, if the subject has not yet been tested on the new phoneme pair, the skill level is set to 1. Testing continues indefinitely, or for the time allotted for Phoneme Identification on the subject's daily training schedule.

Referring now to FIG. 28, a screen shot 2800 is shown for the seventh game in the Fast Forward program, Language Comprehension Builder. The screen shot 2800 contains an ear/hand button 2802 for beginning a trial, a stop sign 2804 for exiting the game, a number score 2806 corresponding to the number of correct responses, and level icons 2808 for indicating the processing level that is currently being tested. In addition, four windows 2810 are shown for containing one to four stimulus images, according to the particular trial being presented. If less than four stimulus images are required for a trial, they are placed randomly within the four windows 2810. At the bottom of the screen 2800 are smaller progress windows 2812 for holding progress elements. The progress elements provide a visual indicator to a subject of his/her progress. As in previously discussed games, when all of the progress elements are obtained, usually ten correct responses, a reward animation is presented to the subject. In one embodiment of this game, the reward animation builds a space ship out of the progress elements.

The stimulus that is provided to the subject is in the form of command sentences. The sentences are divided into 7 comprehension levels, with each level having between 4 to 10 groups of sentences. Each group has 5 sentences. For each stimulus sentence, a corresponding image is provided, with 1-3 distractor images. The subject is to listen to the stimulus sentence and select the corresponding image.

Each of the stimulus sentences may be processed by stretching words, or selected phonemes, in time, and by emphasizing particular frequency envelopes, as shown by table 3040 in FIG. 30. Stretching and emphasis of selected words/phonemes is similar to that described above in other games. The stimulus sentences presented to a subject are provided in Appendix A.

Referring now to FIG. 29, a flow chart 2900 is provided to illustrate the training tutorial aspect of the game. Training begins at block 2902 and proceeds to block 2904.

At block 2904, the subject is prompted to "press the yellow button". That is, the ear/hand button 2802. Flow then proceeds to decision block 2906.

At decision block 2906, a determination is made as to whether the subject has selected the ear/hand button 2802. If not, flow proceeds back to block 2904 where the subject is again prompted, after a predetermined interval. If the subject has pressed the button, flow proceeds to block 2908.

At block 2908, the ear/hand button 2802 is presented, without audio prompting. Flow then proceeds to decision block 2910.

At decision block 2910, a determination is made as to whether the subject has pressed the button 2802. If not, then the subject stays in this loop until the button 2802 is pressed. Once pressed, flow proceeds to block 2912.

At block 2912, a subject is presented with a single image and corresponding audio stimulus. In one embodiment, the stimulus is processed at level 1, with 150% duration and 20 dB selective emphasis. Flow then proceeds to decision block 2914.

At decision block 2914, a determination is made as to whether the subject has selected the image corresponding to the presented stimulus. If not, then flow proceeds back to block 2912 where the subject is again prompted with the stimulus, after a predetermined interval. However, if the subject selected the image, flow proceeds to decision block 2916.

At decision block 2916, a determination is made as to whether the subject has correctly selected an image, 3 times. If not, then flow proceeds back to block 2912 where another image/stimulus combination is presented. However, if the subject has correctly selected an image, 3 times, flow proceeds to block 2918.

At block 2918, an image/stimulus combination is presented, along with a distractor image. Flow then proceeds to decision block 2920.

At decision block 2920, a determination is made as to whether the subject selected the appropriate image. If not, then flow proceeds back to block 2918. However, if the subject selected the correct image, flow proceeds to decision block 2922.

At decision block 2922, a determination is made as to whether the subject has correctly responded to 4 out of 5 trials, on a sliding scale. If not, then flow proceeds back to block 2918. If the subject has correctly responded 4 out of the last 5 trials, flow proceeds to block 2924 allowing the subject to start the game.

Now referring to FIG. 30, a flowchart 3000 is shown illustrating operation of the Language Comprehension Builder game. The game begins at block 3002 and proceeds to block 3004.

At block 3004 an image and stimulus combination is presented to the subject. In one embodiment, the game begins by selecting a group from Level 2, and then by randomly selecting one of the trials from the selected group. The processing of the sentence is performed at 150% duration with 20 dB selected emphasis. Flow then proceeds to decision block 3006.

At decision block 3006, a determination is made as to whether the subject correctly selected the image associated with the stimulus sentence. If not, the subject is shown the correct response, and flow proceeds back to block 3004 where another stimulus/image combination from the same group is presented. If the subject selects the correct image, flow proceeds to decision block 3008.

At decision block 3008, a determination is made as to whether all sentences within a stimulus set have been successfully completed. As mentioned above, the program begins in Level 2, by selecting a particular stimulus set for presentation. The program stays within the selected stimulus set until all stimulus sentences have been responded to correctly. The program then selects another stimulus set from within Level 2. If the subject has not correctly completed all sentences within a stimulus set, flow proceeds back to block 3004 where another sentence is presented. If the subject has completed all stimulus within a set, flow proceeds to decision block 3010.

At decision block 3010, a determination is made as to whether all sets within a particular comprehension level have been completed. If not, then a new set is selected, and flow proceeds back to block 3004. However, if all sets within a comprehension level have been completed, flow proceeds to block 3012.

At block 3012, the comprehension level is incremented. In one embodiment, a subject proceeds through comprehension levels 2-6, in order, with levels 7 and 8 interspersed within levels 3-6. Flow then proceeds to decision block 3014.

At decision block 3014, a determination is made as to whether all comprehension levels have been completed. If not, then flow proceeds back to block 3004 where the subject is presented with an image/stimulus combination from a stimulus set within the new comprehension level. However, if the subject has progressed through all stimulus sets for all comprehension levels, flow proceeds to block 3016.

At block 3016, the processing level applied to the stimulus sets is increased. The processing levels are shown in table 3040. For example, if a subject has just completed processing level 2, having a duration of 125%, and 20 dB emphasis, the processing level is incremented to level 3. This will present all stimulus at 100% duration, and 20 dB emphasis. In addition, it will reset the comprehension level to level 2, and will restart the stimulus set selection. Flow then proceeds to decision block 3018.

At decision block 3018, a determination is made as to whether all processing levels have been completed. If not, then flow proceeds back to block 3004 where a stimulus set from level 2 is presented to the subject, at the new processing level. However, if all the processing levels have been completed, the subject remains at processing level 5 (normal speech). Flow then proceeds to block 3020.

At block 3020, the comprehension levels are reset, so that the subject is presented again with stimulus from level 2. However, no alteration in the stimulus is performed. The subject will remain at processing level 5.

Study has shown that several weeks are required for a subject to advance through all of the comprehension levels, and all of the processing levels. Therefore, when a subject begins each day, he/she is started within the comprehension level, and stimulus set that was last played. And, the stimulus set will be presented at the processing level last played.

In Language Comprehension Builder, as in all of the other games, detailed records are kept regarding each trial, indicating the number of correct responses and incorrect responses, for each processing level, skill level and stimulus set. These records are uploaded to a central server at the end of each day, so that a subject's results may be tabulated and analyzed by an SLP, either working directly with a subject, or remotely. Based on analysis by the SLP, modification to training parameters within Fast ForWord may be made, and downloaded to the subject. This allows a subject to begin each day with a sensory training program that is individually tailored to his/her skill level.

The above discussion provides a detailed understanding of the operation of the present invention as embodied in the game modules within the program entitled Fast ForWord. Each of the game modules present different problems to a subject, using modified phonemes, frequency sweeps or speech commands that are stretched, emphasized or separated in time, according to the subject's ability, and according to predefined processing parameters within the program. Although alternative acoustic processing methodologies may be used, discussion will now be directed at algorithms developed specifically for use by the above described games.

In one embodiment, a two-stage speech modification procedure was used. The first stage involved time-scale modification of speech signals without altering its spectral content. The time scale modification is called the "phase vocoder", and will be further described below. The second speech modification stage that was developed uses an algorithm that differentially amplifies and disambiguates faster phonetic elements in speech. "Fast elements" in speech are defined as those that occur in the 3-30 Hz range within an envelope of narrow-band speech channels of a rate changed speech signal. An emphasis algorithm for these fast elements was implemented using two methods: a filter-bank summation method and an overlap-add method based on a short-time Fourier transform. Both of these emphasis algorithms will be further described below.

Time-scale modification

Referring to FIG. 31, a flow chart 3100 is provided that illustrates time-scale modification of speech signals according to the present invention. Modification begins at block 3102 and proceeds to block 3104.

At block 3104, segmented digital speech input is provided to a processor. The segmented speech is assumed to be broadband and composed of a set of narrow-band signals obtained by passing the speech segment through a filter-bank of band-pass filters. The speech signals may be written as follows: ##EQU1## This is the convolution integral of the signal f(t) and h(t), a prototypical low-pass filter modulated by cos[.omega..sub.n (t)] where .omega..sub.n is the center frequency of the filters in the filter-bank, an operation commonly referred to as heterodyning. Flow then proceeds to block 3106.

At block 3106, the above integral is windowed, and a short-term Fourier transform of the input signal is evaluated at the radian frequency .omega..sub.n using an FFT algorithm. The complex value of this transform is denoted:

where .phi..sub.n (.omega..sub.n,t) is the phase modulation of the carrier cos[on (t)]. Flow then proceeds to block 3108.

At block 3108 the amplitude and phase of the STFT is computed. It is known that the phase function is not a well behaved function, however its derivative, the instanteneous frequency, is bounded and is band limited. Therefore, a practical approximation for f.sub.n (t) is: ##EQU2## where .phi.* is the instantaneous frequency. Flow then proceeds to block 3110.

At block 3110 .PHI.* can be computed from the unwrapped-phase of the short-term Fourier transform. A time-scaled signal can then be synthesized as follows by interpolating the short-term Fourier transform magnitude and the unwrapped phase to the new-time scale as shown below. ##EQU3## where .beta. is the scaling factor which is greater than one for time-scale expansion. An efficient method to compute the above equation makes use of cyclic rotation and the FFT algorithm along with an overlap-add procedure to compute the short-time discrete Fourier transform. Appropriate choice of the analysis filters h(t) and interpolating filters (for interpolation of the short-term Fourier transform to the new time-scale) are important to the algorithm. In one embodiment, linear interpolation based on the magnitude and phase of the short-time Fourier transform was used. The analysis filter h(t) was chosen to be a Kaiser window multiplied by an ideal impulse response as shown: ##EQU4## where I.sub.O (.alpha.) is the zeroth-order modified Bessel function of the first kind and N is the length of the analysis window over which the FFT is computed. Flow then proceeds to block 3112.

At block 3112, a short-term inverse FFT is computed to produce digital speech output. This output is then provided at block 3114.

Filter -bank emphasis algorithm

Now referring to FIG. 32, a flow chart 3200 is shown that illustrates implementation of an emphasis algorithm according to the present invention. The algorithm begins at block 3202 and proceeds to block 3204.

At block 3204, it is assumed that the speech signal can be synthesized through a bank of band-pass filters, as described above. This time, however, no heterodyning of a prototypical low-pass filter is used. Instead, a set of up to 20 second-order Butterworth filters with center frequencies logarithmically spaced between 100 and the nyquist frequency are used. The output of each band-pass filter resulted in a narrow-band channel signal f.sub.n (t). Flow then proceeds to block 3206.

At block 3206, we computed the analytical signal as follows:

where H(n) is the Hilbert transform of a signal defined as: ##EQU5##

The Hilbert transform was computed using the FFT algorithm. It is known that the absolute value of the analytical signal is the envelope of a narrow-band signal. Thus, an envelope e.sub.n (n) is obtained by the following operation:

The envelope within each narrow-band channel is then band-pass filtered using a second order Butterworth filter with the cutt-offs set usually between 3-30 Hz (the time scale at which phonetic events occur in rate changed speech). The band pass filtered envelope is then rectified to form the new envelope as follows:

where

and g(n) is the impulse-response of the band-pass second order Butterworth filter. Flow then proceeds to block 3208.

At block 3208, the signal is modified within each band-pass channel to carry this new envelope, as shown below: ##EQU6## Flow then proceeds to block 3210.

At block 3210 the modified signal is obtained by summing the narrow-band filters with a differential gain for each channel as follows: ##EQU7## where w.sub.n is the gain for each channel. The envelope is modified only within a specified frequency range from 1-10 KHz which normally spans about 16 channels. Flow then proceeds to block 3212.

At block 3212 segmented digital speech output is provided.

Overlap-add emphasis algorithm

Referring to FIG. 33, a flow chart 3300 for an alternative emphasis algorithm is provided. This algorithm improves upon the filter-bank summation described above by making use of the property of equivalence between the short-time Fourier transform and the filter-bank summation algorithm. In this embodiment, the short-time Fourier transform is computed using an overlap-add procedure and the FFT algorithm. Flow begins at block 3302 and proceeds to block 3304.

At block 3304, the short-time Fourier transform is computed over a sliding window given by the following equation: ##EQU8## where h(n) is a Hamming window and the overlap between sections was chosen to be less than a quarter the length of the analysis window. The envelope can then be obtained within narrow-band channels from the absolute value of the short-time Fourier transform. The number of narrow-band channels is equal to half the size of the length over which the FFT is computed.

The energy of the envelope within critical band channels is then averaged, as shown: ##EQU9## where C.sub.n is the corner-frequency of the critical-band channel n. At present, critical-band frequencies for children with LLI are unknown, therefore the present invention approximates the bands using parameters proposed by Zwicker. See E. Zwicker and E. Terhardt, "Analytical expressions for critical-band rate and critical bandwidth as a function of frequency," J. Acoust. Soc. Ame., vol. 68, pp. 1523-25, 1980. As critical band frequencies for children with LLI become available, they can be incorporated into the present invention.

The envelope within each critical-band channel is then band-pass-filtered with cut off's set usually between 3-30 Hz with type I linear phase FIR equiripple filters. The band-pass filtered envelope is then threshold rectified. In contrast to the filter-bank emphasis algorithm, the modified envelope is added to the original envelope to amplify the fast elements while not distorting the slower modulations. This is given by the following equation: ##EQU10## where,

Flow then proceeds to block 3308.

At block 3308, a modified signal is obtained by summing the short-time Fourier transform using a weighted overlap-add procedure as shown below: ##EQU11## where g(n) is the synthesis filter which was also chosen to be a Hamming window. Flow then proceeds to block 3310. At block 3310, windowing and over-lap addition for the algorithm is performed. Flow then proceeds to block 3312 where segmented digital speech output is provided.

Although the present invention and its objects, features, and advantages have been described in detail, other embodiments are encompassed by the invention. For example, a number of different games have been shown, each dealing with a stimulus set that is processed in the time domain, and presented to a subject in a manner that the subject can understand. The processing is designed to emphasize or stretch those components of speech that are the most difficult for an LLI subject to differentiate, so that they may be more easily understood. In addition, the processing allows distinct frequency sweeps, phonemes or words to be separated in time, first by a significant amount, say 500 ms, with the amount of separation being gradually reduced to that of normal speech. Once a subject gains success at distinguishing between similar speech elements, at a high level of processing, the amount of processing is gradually reduced until it reaches the level of normal speech. The particular games used to train subjects have distinct advantages, but are not exclusive. Other games are anticipated that will incorporate the novel aspects of the present invention, to further train a subject's temporal processing ability to recognize and distinguish between short duration acoustic events that are common in speech.

Furthermore, the Fast ForWord program has been shown for execution on a personal computer, connected to a central server. However, as technology advances, it is envisioned that the program could be executed either by a diskless computer attached to a server, by a handheld processing device, such as a laptop, or eventually by a palmtop device such as a Nintendo GameBoy. As long as the graphical images and auditory prompts can be presented in a timely fashion, and with high quality, the nature of the device used to present the material is irrelevant.

Those skilled in the art should appreciate that they can readily use the disclosed conception and specific embodiments as a basis for designing or modifying other structures for carrying out the same purposes of the present invention without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

We claim:

1. A computerized method for improving a subject's comprehension of logical constructs within language, using speech commands that are modified by computer processing, the method comprising: a) providing a plurality of training skill levels for the speech commands, the plurality of training skill levels differing from each other in the difficulty of logical constructs, and in the processing applied to the speech commands by a computer; b) selecting from the plurality of training skill levels, a training skill level for presentation to the subject that is associated with the subject's ability; c) presenting via a computer display a game board having a plurality of graphical components, the plurality of graphical components being the subject of the speech commands modified by the computer; d) presenting via the computer as a trial a speech command from a set of the speech commands modified by the computer from the selected skill level, the speech command directing the subject to manipulate at least one of the plurality of graphical components presented via the computer display; e) repeating said step of presenting as a trial, if the subject correctly manipulates the at least one of the graphical components; and f) visually indicating to the subject the correct manipulation, if the subject incorrectly manipulates the at least one of the graphical components.

2. The computerized method as recited in claim 1 wherein the computer processing applied to the speech commands comprises: stretching the speech commands in the time domain without significantly affecting frequency characteristics of the speech commands; and emphasizing frequency envelopes within a consonant portion of the speech commands.

3. The computerized method as recited in claim 2 wherein the plurality of training skill levels differ from each other with respect to the amount of stretching and emphasis applied to the speech commands.

4. The computerized method as recited in claim 1 further comprising: applying the logical constructs to the speech commands to direct the subject to recognize at least two characteristics of one of the plurality of graphical components.

5. The computerized method as recited in claim 4 wherein the characteristics of the one of the plurality of graphical component comprises: color; size; and shape.

6. The computerized method as recited in claim 2 wherein a first skill level within the plurality of training skill levels comprises: a speech command directing the subject to recognize at least two characteristics of one of the plurality of graphical components, wherein the speech command is stretched, in the time domain, by at least 125%, and frequency envelopes within a consonant portion of the speech command are emphasized by at least 10 dB.

7. The computerized method as recited in claim 2 further comprising: g) recording correct/incorrect manipulations of the graphical components by the subject.

8. The computerized method as recited in claim 7 wherein the manipulations recorded by said recording are used by said step of selecting to select a training skill level.

9. The computerized method as recited in claim 8 further comprising: selecting another skill level from the plurality of training skill levels, that increases the difficulty of the trial, as the subject correctly responds to a plurality of trials.

10. The computerized method as recited in claim 9 wherein the difficulty of the trial is increased by reducing the amount of computer processing applied to the speech commands.

11. The computerized method as recited in claim 8 further comprising: selecting another skill level from the plurality of training skill levels, that decreases the difficulty of the trial, as the subject incorrectly responds to a plurality of trials.

12. The computerized method as recited in claim 11 wherein the difficulty of the trial is decreased by said step of selecting by increasing the computer processing applied to the speech commands.

13. A computerized method for improving the cognitive processing system of a subject, the method presented to the subject on a computer, and interacting with the subject via input/output devices on the computer, the method comprising: a) providing a plurality of stimulus sets, each of the plurality of stimulus sets comprising a plurality of aural command sentences; b) providing a plurality of target graphical images, each of the graphical images associated with a different one of the plurality of command sentences; c) providing a plurality of distractor images, not associated with the plurality of command sentences; d) presenting to the subject, via the computer, one of the plurality of command sentences from one of the plurality of stimulus sets to the subject, the presented sentence modified acoustically by the computer; e) presenting to the subject on a computer display, a target graphical image, from the plurality of target graphical images, that is associated with the presented command sentence; f) presenting to the subject, along with said step e) a plurality of distractor images; and g) requiring the subject to distinguish between the presented target graphical image, and the presented plurality of distractor images by selecting the target graphical image associated with the presented command sentence.

14. The method as recited in claim 13 wherein each of the plurality of stimulus sets group the plurality of command sentences according to the difficulty of the language constructs.

15. The method as recited in claim 13 wherein said step a) further comprises: providing a plurality of skill levels for each of the plurality of command sentences.

16. The method as recited in claim 15 wherein said step of providing a plurality of skill levels comprises: providing a plurality of modified command sentences, corresponding to each of the plurality of skill levels, for each of the plurality of command sentences.

17. The method as recited in claim 16 wherein the modified command sentences differ from each other in the amount of stretching and emphasis applied to the command sentences.

18. The method as recited in claim 17 wherein each of the plurality of modified command sentences are stretched by the computer in the time domain, between 100% and approximately 150%.

19. The method as recited in claim 17 wherein each of the plurality of modified command sentences has frequency envelopes within consonant portions of the sentences that are emphasized between 0 dB and 20 dB.

Patent Citations (68)

Patent	Date	Inventor
US3816664	1974-06-01	Koch
US3920903	1975-11-01	Beller
US4010557	1977-03-01	Martin et al.
US4128737	1978-12-01	Dorais
US4327252	1982-04-01	Tomatis
US4372558	1983-02-01	Shimamoto et al.
US4464119	1984-08-01	Vildgrude et al.
US4586905	1986-05-01	Groff
US4593904	1986-06-01	Graves
US4641343	1987-02-01	Holland et al.
US4696042	1987-09-01	Goudie
US4708657	1987-11-01	Beller et al.
US4736429	1988-04-01	Niyada et al.
US4741037	1988-04-01	Goldstern
US4799261	1989-01-01	Lin et al.
US4813076	1989-03-01	Miller
US4820059	1989-04-01	Miller et al.
US4821325	1989-04-01	Martin et al.
US4852168	1989-07-01	Sprague
US4852170	1989-07-01	Bordeaux
US4867685	1989-09-01	Brush et al.
US4884972	1989-12-01	Gasper
US4895519	1990-01-01	Beller et al.
US4907274	1990-03-01	Nomura et al.
US4980917	1990-12-01	Hutchins
US5010495	1991-04-01	Willetts
US5046735	1991-09-01	Hamano et al.
US5056143	1991-10-01	Taguchi
US5119826	1992-06-01	Baart de la Faille
US5123048	1992-06-01	Miyamae et al.
US5169342	1992-12-01	Steele et al.
US5180309	1993-01-01	Egnor
US5211564	1993-05-01	Martinez et al.
US5215468	1993-06-01	Lauffer et al.
US5285498	1994-02-01	Johnston
US5289521	1994-02-01	Coleman et al.
US5295223	1994-03-01	Saito
US5296845	1994-03-01	Haller
US5302132	1994-04-01	Corder
US5305420	1994-04-01	Nakamura et al.
US5340316	1994-08-01	Javkin et al.
US5341432	1994-08-01	Suzuki et al.
US5387104	1995-02-01	Corder
US5393236	1995-02-01	Blackmer et al.
US5411271	1995-05-01	Mirando
US5421731	1995-06-01	Walker
US5423556	1995-06-01	Latypov
US5429513	1995-07-01	Diaz-Plaza
US5456607	1995-10-01	Antoniak
US5487671	1996-01-01	Shpiro et al.
US5517595	1996-05-01	Kleijn
US5536171	1996-07-01	Javkin et al.
US5540589	1996-07-01	Waters
US5562453	1996-10-01	Wen
US5573403	1996-11-01	Beller et al.
US5577185	1996-11-01	Tunnell et al.
US5618182	1997-04-01	Thomas
US5630753	1997-05-01	Fuchs
US5678001	1997-10-01	Nagel et al.
US5690493	1997-11-01	McAlear, Jr.
US5697789	1997-12-01	Sameth et al.
US5787234	1998-07-01	Molloy
US5882258	1999-03-01	Kelly et al.
US5927988	1999-07-01	Jenkins et al.
EPX534410A2	1993-03-01
EPX561752A1	1993-09-01
FRX2695750	1994-03-01
WOXWO96/18184	1996-06-01

Non-Patent Literature (27)

Alexander, A., et al., "Phonological Awareness Training and Remediation of Analytic Decoding Deficits in a Group of Severe Dyslexics," Annals of Dyslexia, 41:193-206 (1991).
Alexander, D. and Frost, B. "Decelerated Synthesized Speech as a Means of Shaping Speed of Auditory Processing of Children with Delayed Language", Perceptual and Motor Skills, pp. 783-792, 1982, vol. 55.
Barinaga, M., "Giving Language Skills a Boost", Jan. 5, 1996, Science, vol. 271, p. 27-28.
Berard, G., "Hearing Equals Behavior", pp. 77-93, (Keats Publishing, Inc. 1993).
Berkell, D. et al "Auditory Integration Training for Individuals with Autism,: Education and Training in Mental Retardation and Developmental Disabilities." 31, n. 1:66-70 (1996).
Burne, J.,, "Crazy sounds to soothe a garbled world," The Independent II, p. 24 (1994).
Elliott, L., Hammer, M., Schol, M., Carrell, T., and Wasowicz, J., "Discrimination of Rising and Falling Simulated Single-Formant Frequency Transitions: Practice and Transition Duration Effects," J. Acoust. Soc. Am. 86(3), 1989.
Fiez, J., "PET Activation of Posterior Temporal Regions During Auditory Word Presentation and Verb Generation", Jan./Feb. 1996, Cerebral Cortex, 6:1-9.
Fiez, J., "PET Studies of Auditory and Phonological Processing: Effects of Stimulus Characteristics and Task Demands", 1995, Journal of Cognitive Neuroscience 7:3, p. 357-375.
Flowers, A., "Auditory Perception: Speech, Language and Learning," Chapter 9, Perceptual Learning Systems, Dearborn, Michigan, 1983, p. 133-138.
International Dyslexia Organization web site, www.interdys.org (1999).
Levinson, P. and Sloan, C., Auditory Processing and Language: Clinical and Research Perspectives, Chapter 7, Grune & Stratton, New York, New York (1980), p. 117-133.
Mehler, J., et al., "Understanding Compressed Sentences: The Role of Rhythm and Meaning," Annals of the New York Academy of Sciences, 682:272-282 (1993).
Merzenich, M., et al. "Cortical Plasticity Underlying Perceptual, Motor, and Cognitive Skill Development: Implications for Neurorehabilitation", 1996, Cold Spring Habor Symposium on Quant. Bio. vol. LXI, p. 1-8.
Merzenich, M., et al. "Cortical Plasticity, Learning, and Language Dysfunction", 1995, SFI Studies in Science of Complexity, vol. XXIII, p. 247-272.
Merzenich, M., "Temporal Processing Deficits of Language-Learning Impaired Children Ameliorated by Training", Jan. 5, 1996, Science, vol. 271, p. 77-81.
Rimland, B. et al. "Brief Report: A Pilot Study of Auditory Integration Training in Autism," Journal of Autism and Developmental Disorders, 25, n. 1: 61-70 (1995).
Rimland, B., "Summaries of Research on Auditory Integration Training" (1993-1995: 11 Studies), Autism Research Institute (1995).
Sloan, C., "Treating Auditory Processing Difficulties in Children", Chapters 4, 5, 7, Singular Publishing Group, Inc., San Diego, CA (1986; reprinted 1991), p. 35-55, 57-61 and 63-82.
Tallal, P., et at., "Language Comprehension in Language-Learning Impaired Children Improved with Acoustically Modified Speech," Science, 271:81 1996.
Tallal, P., "Implications of Speech Perceptual Research for Clinical Populations", in Kavanagh, James F., "Speech and Language in the Laboratory, School, and Clinic" (1978), p. 73-88.
Tallal, P., et al., "Developmental Aphasia: Rate of Auditory Processing and Selective Impairment of Consonant Perception," Neuropsychologia, 12:83-93, 1974.
Tallal, P., et al., "Developmental Aphasia: The Perception of Brief Vowels and Extended Stop Consonants," Neuropsychologia, 13:69-74 (1975).
Tallal, P., et al., "Neurobiological Basis of Speech: A Case for the Preeminence of Temporal Processing," Annals. of the New York Academy of Sciences,.682:2.7-47 (1993).
Tallal, P., et al., "The Role of Temporal Processing in Developmental Language-Based Learning Disorders: Research and Clinical Implications", Ch. 7, Blachman's Foundations of Reading Acquisition and Dyslexia, 1997.
Tallal, P., et al.., "Defects 'of Non-Verbal Auditory Perception in Children with Developmental Aphasia," Nature, 241: 468-469 (1973).
Tallal, P., et al.., "Developmental Aphasia: Impaired Rate of Non-Verbal Processing as a Function of Sensory Modality," Neuropsychologia, 11:389-398 (1973).

Abstract

Description