DYNAMIC THRESHOLDS FOR ALWAYS LISTENING SPEECH TRIGGER

[0001] This application claims priority from U.S. Provisional Ser. No. 62/049,917, filed on Sep. 12, 2014, entitled DYNAMIC THRESHOLDS FOR ALWAYS LISTENING SPEECH TRIGGER, which is hereby incorporated by reference in its entirety for all purposes.

[0002] This relates generally to speech triggers and, more specifically, to dynamically adjusting always listening speech trigger thresholds.

[0003] Intelligent automated assistants (or virtual assistants) provide an intuitive interface between users and electronic devices. These assistants can allow users to interact with devices or systems using natural language in spoken and/or text forms. For example, a user can access the services of an electronic device by providing a spoken user input in natural language form to a virtual assistant associated with the electronic device. The virtual assistant can perform natural language processing on the spoken user input to infer the user's intent and operationalize the user's intent into tasks. The tasks can then be performed by executing one or more functions of the electronic device, and a relevant output can be returned to the user in natural language form.

[0004] Virtual assistant interactions can be triggered in a variety of ways. In one example, a virtual assistant session can be initiated in response to a user pressing a physical or virtual button on a user device, after which the virtual assistant receives a user command, determines the user's intent, and executes the associated command. In other examples, virtual assistant sessions can be initiated in response to detecting a spoken trigger. For example, a user device can listen for a trigger word or phrase at all times (e.g., always listening), in certain situations, at particular times, or the like. Upon detecting the spoken trigger word or phrase, a virtual assistant session can be initiated in which the virtual assistant receives a user command, determines the user's intent, and executes the associated command. Spoken triggers can also be used to cause others actions and initiate interactions with a variety of other systems and devices.

[0005] Speech triggers, however, can be missed due to background noise, interference, variations in user speech, and a variety of other factors. For example, a user may utter a speech trigger softly, in a noisy environment, in a unique tone of voice, or the like, and a virtual assistant may not be triggered. Users in such instances may repeat the speech trigger louder or more clearly, or they may resort to manually initiating a virtual assistant session. Missed triggers can thus be frustrating for users. In addition to being missed, speech triggers can also be falsely recognized leading to false positive triggering events that impair the user experience, waste device power, and the like. For example, background speech from a nearby person can be detected and can lead undesirably to a false positive triggering event that a user never intended.

[0006] In some examples, a threshold confidence level can be used to determine whether to trigger a virtual assistant based on sampled audio that may contain a spoken trigger. A confidence score or level can be determined based on the likelihood that the audio includes an intentionally spoken trigger word or phrase. A virtual assistant session can be initiated depending on whether the confidence score is above or below a predetermined threshold. Confidence scores below the threshold may not trigger the virtual assistant, while confidence scores above the threshold may trigger the virtual assistant. As environmental factors and usage vary, however, a predetermined threshold can undesirably yield missed triggers and false positive triggering events.

[0007] Accordingly, always listening speech triggers with predetermined triggering thresholds can suffer poor accuracy as environmental and usage factors vary, which can limit utility and negatively impact the user experience.

[0008] Systems and processes are disclosed for dynamically adjusting a speech trigger threshold. In one example, audio input can be received via a microphone. The audio input can be sampled. A confidence level can be determined that the sampled audio input includes a portion of a spoken trigger. In response to the confidence level exceeding a threshold, a virtual assistant can be triggered to receive a user command from the audio input. The threshold can be dynamically adjusted (e.g., raised or lowered) in response to a variety of perceived events. In some examples, triggering the virtual assistant includes initiating a virtual assistant session with a user. In other examples, a user intent can be determined based on the received user command, and a command associated with the determined user intent can be executed.

[0016] In the following description of examples, reference is made to the accompanying drawings in which it is shown by way of illustration specific examples that can be practiced. It is to be understood that other examples can be used and structural changes can be made without departing from the scope of the various examples.

[0017] This relates to systems and processes for dynamically adjusting a speech trigger threshold, which can be used, for example, in triggering a virtual assistant or initiating other actions or interactions. In one example, accurate speech triggering can be achieved by dynamically adjusting a speech trigger threshold based on perceived events (e.g., environmental factors, device usage, etc.). The speech trigger threshold can be lowered, for example, when perceived events indicate a user may be more likely to initiate a virtual assistant session using a speech trigger or a speech trigger may be more difficult to detect. This can minimize missed triggers by increasing the likelihood of triggering. Conversely, the speech trigger threshold can be raised, for example, when perceived events indicate a user may be less likely to initiate a virtual assistant session using a speech trigger. This can minimize false positive triggering events by decreasing the likelihood of triggering.

[0018] In one example, audio input can be received via a microphone (e.g., of a user device). The received audio input can be sampled, and a confidence level can be determined of whether the sampled audio input includes a portion of a spoken trigger. In response to the confidence level exceeding a threshold, a virtual assistant can be triggered to receive a user command from the audio input. The threshold can be dynamically adjusted in response to perceived events, thereby minimizing both missed triggers and false positive triggering events.

[0019] It should be understood that dynamically adjusting a speech trigger threshold can provide accurate speech triggering, significant system utility, and an enjoyable user experience. Device power can also be conserved by minimizing false positive triggering events and thereby limiting undesired virtual assistant sessions or other triggered actions. It should be understood, however, that still many other advantages can be achieved according to the various examples discussed herein.

[0020] FIG. 1 illustrates exemplary system 100 for dynamically adjusting a speech trigger threshold and triggering a virtual assistant according to various examples. It should be understood that adjusting a speech trigger threshold as discussed herein can be used for any of a variety of applications, including in support of a virtual assistant. In other examples, speech trigger thresholds can be used to cause a variety of other actions and to initiate interactions with any of a variety of other systems and devices, and such speech trigger thresholds can be dynamically adjusted according to the various examples herein. The terms “virtual assistant,” “digital assistant,” “intelligent automated assistant,” or “automatic digital assistant” can refer to any information processing system that can interpret natural language input in spoken and/or textual form to infer user intent, and perform actions based on the inferred user intent. For example, to act on an inferred user intent, the system can perform one or more of the following: identifying a task flow with steps and parameters designed to accomplish the inferred user intent; inputting specific requirements from the inferred user intent into the task flow; executing the task flow by invoking programs, methods, services, APIs, or the like; and generating output responses to the user in an audible (e.g., speech) and/or visual form.

[0021] A virtual assistant can be capable of accepting a user request at least partially in the form of a natural language command, request, statement, narrative, and/or inquiry. Typically, the user request seeks either an informational answer or performance of a task by the virtual assistant. A satisfactory response to the user request can include provision of the requested informational answer, performance of the requested task, or a combination of the two. For example, a user can ask the virtual assistant a question, such as “Where am I right now?” Based on the user's current location, the virtual assistant can answer, “You are in Central Park.” The user can also request the performance of a task, for example, “Please remind me to call Mom at 4 p.m. today.” In response, the virtual assistant can acknowledge the request and then create an appropriate reminder item in the user's electronic schedule. During the performance of a requested task, the virtual assistant can sometimes interact with the user in a continuous dialogue involving multiple exchanges of information over an extended period of time. There are numerous other ways of interacting with a virtual assistant to request information or performance of various tasks. In addition to providing verbal responses and taking programmed actions, the virtual assistant can also provide responses in other visual or audio forms (e.g., as text, alerts, music, videos, animations, etc.).

[0022] An example of a virtual assistant is described in Applicants' U.S. Utility Application Ser. No. 12/987,982 for “Intelligent Automated Assistant,” filed Jan. 10, 2011, the entire disclosure of which is incorporated herein by reference.

[0023] As shown in FIG. 1, in some examples, a virtual assistant can be implemented according to a client-server model. The virtual assistant can include a client-side portion executed on a user device 102, and a server-side portion executed on a server system 110. User device 102 can include any electronic device, such as a mobile phone (e.g., smartphone), tablet computer, portable media player, desktop computer, laptop computer, PDA, television, television set-top box (e.g., cable box, video player, video streaming device, etc.), wearable electronic device (e.g., digital glasses, wristband, wristwatch, brooch, armband, etc.), gaming system, home security system, home automation system, vehicle control system, or the like. User device 102 can communicate with server system 110 through one or more networks 108, which can include the Internet, an intranet, or any other wired or wireless public or private network.

[0024] The client-side portion executed on user device 102 can provide client-side functionalities, such as user-facing input and output processing and communications with server system 110. Server system 110 can provide server-side functionalities for any number of clients residing on a respective user device 102.

[0025] Server system 110 can include one or more virtual assistant servers 114 that can include a client-facing I/O interface 122, one or more processing modules 118, data and model storage 120, and an I/O interface to external services 116. The client-facing I/O interface 122 can facilitate the client-facing input and output processing for virtual assistant server 114. The one or more processing modules 118 can utilize data and model storage 120 to determine the user's intent based on natural language input, and can perform task execution based on inferred user intent. In some examples, virtual assistant server 114 can communicate with external services 124, such as telephony services, calendar services, information services, messaging services, navigation services, and the like, through network(s) 108 for task completion or information acquisition. The I/O interface to external services 116 can facilitate such communications.

[0026] Server system 110 can be implemented on one or more standalone data processing devices or a distributed network of computers. In some examples, server system 110 can employ various virtual devices and/or services of third party service providers (e.g., third-party cloud service providers) to provide the underlying computing resources and/or infrastructure resources of server system 110.

[0027] Although the functionality of the virtual assistant is shown in FIG. 1 as including both a client-side portion and a server-side portion, in some examples, the functions of an assistant (or speech recognition in general) can be implemented as a standalone application installed on a user device. In addition, the division of functionalities between the client and server portions of the virtual assistant can vary in different examples. For instance, in some examples, the client executed on user device 102 can be a thin-client that provides only user-facing input and output processing functions, and delegates all other functionalities of the virtual assistant to a backend server.

[0028] FIG. 2 illustrates a block diagram of exemplary user device 102 according to various examples. As shown, user device 102 can include a memory interface 202, one or more processors 204, and a peripherals interface 206. The various components in user device 102 can be coupled together by one or more communication buses or signal lines. User device 102 can further include various sensors, subsystems, and peripheral devices that are coupled to the peripherals interface 206. The sensors, subsystems, and peripheral devices can gather information and/or facilitate various functionalities of user device 102.

[0029] For example, user device 102 can include a motion sensor 210, a light sensor 212, and a proximity sensor 214 coupled to peripherals interface 206 to facilitate orientation, light, and proximity sensing functions. One or more other sensors 216, such as a positioning system (e.g., a GPS receiver), a temperature sensor, a biometric sensor, a gyroscope, a compass, an accelerometer, and the like, can also be connected to peripherals interface 206, to facilitate related functionalities.

[0030] In some examples, a camera subsystem 220 and an optical sensor 222 can be utilized to facilitate camera functions, such as taking photographs and recording video clips. Communication functions can be facilitated through one or more wired and/or wireless communication subsystems 224, which can include various communication ports, radio frequency receivers and transmitters, and/or optical (e.g., infrared) receivers and transmitters. An audio subsystem 226 can be coupled to speakers 228 and microphone 230 to facilitate voice-enabled functions, such as speech trigger detection, voice recognition, voice replication, digital recording, and telephony functions.

[0031] In some examples, user device 102 can further include an I/O subsystem 240 coupled to peripherals interface 206. I/O subsystem 240 can include a touchscreen controller 242 and/or other input controller(s) 244. Touchscreen controller 242 can be coupled to a touchscreen 246. Touchscreen 246 and the touchscreen controller 242 can, for example, detect contact and movement or break thereof using any of a plurality of touch sensitivity technologies, such as capacitive, resistive, infrared, and surface acoustic wave technologies, proximity sensor arrays, and the like. Other input controller(s) 244 can be coupled to other input/control devices 248, such as one or more buttons, rocker switches, a thumb-wheel, an infrared port, a USB port, and/or a pointer device, such as a stylus.

[0032] In some examples, user device 102 can further include a memory interface 202 coupled to memory 250. Memory 250 can include any electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, a portable computer diskette (magnetic), a random access memory (RAM) (magnetic), a read-only memory (ROM) (magnetic), an erasable programmable read-only memory (EPROM) (magnetic), a portable optical disc such as CD, CD-R, CD-RW, DVD, DVD-R, or DVD-RW, or flash memory such as compact flash cards, secured digital cards, USB memory devices, memory sticks, and the like. In some examples, a non-transitory computer-readable storage medium of memory 250 can be used to store instructions (e.g., for performing some or all of process 300, described below) for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device, and can execute the instructions. In other examples, the instructions (e.g., for performing process 300, described below) can be stored on a non-transitory computer-readable storage medium of server system 110, or can be divided between the non-transitory computer-readable storage medium of memory 250 and the non-transitory computer-readable storage medium of server system 110. In the context of this document, a “non-transitory computer readable storage medium” can be any medium that can contain or store the program for use by or in connection with the instruction execution system, apparatus, or device.

[0033] In some examples, memory 250 can store an operating system 252, a communication module 254, a graphical user interface module 256, a sensor processing module 258, a phone module 260, and applications 262. Operating system 252 can include instructions for handling basic system services and for performing hardware dependent tasks. Communication module 254 can facilitate communicating with one or more additional devices, one or more computers, and/or one or more servers. Graphical user interface module 256 can facilitate graphic user interface processing. Sensor processing module 258 can facilitate sensor related processing and functions. Phone module 260 can facilitate phone-related processes and functions. Application module 262 can facilitate various functionalities of user applications, such as electronic messaging, web browsing, media processing, navigation, imaging, and/or other processes and functions.

[0034] As described herein, memory 250 can also store client-side virtual assistant instructions (e.g., in a virtual assistant client module 264) and various user data 266 (e.g., user-specific vocabulary data, preference data, and/or other data such as the user's electronic address book, to-do lists, shopping lists, etc.) to, for example, provide the client-side functionalities of the virtual assistant. User data 266 can also be used in performing speech recognition in support of the virtual assistant or for any other application.

[0035] In various examples, virtual assistant client module 264 can be capable of accepting voice input (e.g., speech input), text input, touch input, and/or gestural input through various user interfaces (e.g., I/O subsystem 240, audio subsystem 226, or the like) of user device 102. Virtual assistant client module 264 can also be capable of providing output in audio (e.g., speech output), visual, and/or tactile forms. For example, output can be provided as voice, sound, alerts, text messages, menus, graphics, videos, animations, vibrations, and/or combinations of two or more of the above. During operation, virtual assistant client module 264 can communicate with the virtual assistant server using communication subsystem 224.

[0036] In some examples, virtual assistant client module 264 can utilize the various sensors, subsystems, and peripheral devices to gather additional information from the surrounding environment of user device 102 to establish a context associated with a user, the current user interaction, and/or the current user input. In some examples, virtual assistant client module 264 can provide the contextual information or a subset thereof with the user input to the virtual assistant server to help infer the user's intent. The virtual assistant can also use the contextual information to determine how to prepare and deliver outputs to the user. The contextual information can further be used by user device 102 or server system 110 to support accurate speech recognition, as discussed herein.

[0037] In some examples, the contextual information that accompanies the user input can include sensor information, such as lighting, ambient noise, ambient temperature, images or videos of the surrounding environment, distance to another object, and the like. The contextual information can further include information associated with the physical state of user device 102 (e.g., device orientation, device location, device temperature, power level, speed, acceleration, motion patterns, cellular signal strength, etc.) or the software state of user device 102 (e.g., running processes, installed programs, past and present network activities, background services, error logs, resources usage, etc.). Any of these types of contextual information can be provided to virtual assistant server 114 (or used on user device 102 itself) as contextual information associated with a user input.

[0038] In some examples, virtual assistant client module 264 can selectively provide information (e.g., user data 266) stored on user device 102 in response to requests from virtual assistant server 114 (or it can be used on user device 102 itself in executing speech recognition and/or virtual assistant functions). Virtual assistant client module 264 can also elicit additional input from the user via a natural language dialogue or other user interfaces upon request by virtual assistant server 114. Virtual assistant client module 264 can pass the additional input to virtual assistant server 114 to help virtual assistant server 114 in intent inference and/or fulfillment of the user's intent expressed in the user request.

[0039] In various examples, memory 250 can include additional instructions or fewer instructions. Furthermore, various functions of user device 102 can be implemented in hardware and/or in firmware, including in one or more signal processing and/or application specific integrated circuits.

[0040] It should be understood that system 100 is not limited to the components and configuration shown in FIG. 1, and user device 102 is likewise not limited to the components and configuration shown in FIG. 2. Both system 100 and user device 102 can include fewer or other components in multiple configurations according to various examples.

[0041] FIG. 3 illustrates exemplary process 300 for dynamically adjusting a speech trigger threshold and triggering a virtual assistant. Process 300 can, for example, be executed on processor 204 of user device 102 discussed above with reference to FIG. 2. In other examples, processing modules 118 of server system 110 and processor 204 of user device 102 can be used together to execute some or all of process 300. At block 302, audio input can be received via a microphone, such as microphone 230 of user device 102. At block 304, the received audio input can be sampled. In one example, received audio input can be sampled at regular intervals, such as sampling a portion of the received audio every ten milliseconds. In other examples, received audio input can be continuously sampled or analyzed at various other intervals. It should be understood that always listening can include receiving and analyzing audio continuously or can include receiving and analyzing audio at regular intervals or at particular times.

[0042] At block 306, a confidence level can be determined that the sampled audio input comprises a portion of a spoken trigger. In one example, sampled portions of audio can be analyzed to determine whether they include portions of a speech trigger. Speech triggers can include any of a variety of spoken words or phrases that a user can utter to trigger an action (e.g., initiating a virtual assistant session). For example, a user can utter “hey Siri” to initiate a session with a virtual assistant referred to as “Siri.” In other examples, a user can utter a designated assistant name or reference, such as “Assistant,” “Siri,” “Secretary,” “Hi Assistant,” “Hello Helper,” or any other names or references.

[0043] In other examples, spoken triggers can include commands, actions, queries, or other actionable words or phrases. For example, certain commands or questions can be available at all times or in certain situations, and those commands or questions can be used as spoken triggers to initiate actions (e.g., executing the associated commands or responding to questions). In some examples, application or program names can also be used as spoken triggers to launch those applications. A user can utter, for example, “Power on,” “Send a message,” “What time is it,” “What's my next meeting,” “Launch the camera application,” “Camera,” “Email,” “Messages,” “Play music,” “Pause music,” “Radio,” “Navigation,” “Calculator,” “Notes,” “Flashlight,” “Phone,” or any of a variety of other application names, commands, and queries that can be recognized as spoken triggers to launch an application, execute a command, or respond to a query. In some examples, launching some applications, executing certain commands, and responding to certain queries can be done without fully initiating a session with a virtual assistant (e.g., without initiating virtual assistant interactions). In other examples, a virtual assistant session can be initiated in response to such utterances and the associated functions can be executed without further spoken input from the user. In still other examples, a user can utter a trigger phrase and a command or question in sequence without waiting for confirmation or prompting in between (e.g., “Hey Siri, what's the weather today?”), and the virtual assistant can execute the command or respond to the question without issuing a prompt or implementing other delays in between the trigger phrase and the command or question.

[0044] It should thus be understood that a variety of words, phrases, actions, commands, queries, combinations of such, and the like can be used as spoken triggers. Likewise, although the example of triggering a virtual assistant (e.g., initiating a session) is described herein, spoken triggers can be used for a variety of others actions, commands, queries, functions, and the like, and the examples described herein of dynamically adjusting trigger thresholds can be used for any other spoken triggers associated with any other functions.

[0045] A confidence level or score for the likelihood that sampled audio includes a portion of a spoken trigger can be determined at block 306 in a variety of ways. In one example, sampled audio can be compared to audio files or other digital representations of accepted speech triggers, and a score or level can be determined based on how well the sampled audio matches the speech trigger representation. In other examples, a sequence of phonemes can be determined from sampled audio, and the phoneme sequence can be compared to a phoneme sequence of an accepted speech trigger. A score or level can then be determined based on how well the phoneme sequence of the sampled audio matches the phoneme sequence of the accepted speech trigger. In still other examples, sampled audio can be compared to portions of a spoken trigger in a variety of other ways to determine a score or level for how well the sampled audio matches a portion of a spoken trigger. One of ordinary skill in the art will appreciate that there are a variety of approaches that can be used to determine the likelihood (e.g., confidence level, confidence score, etc.) that an audio sample contains or corresponds to a portion or all of a speech trigger.

[0046] At decision block 308, a determination can be made as to whether the confidence level determined at block 306 exceeds a threshold. In one example, a speech trigger confidence level threshold can be predetermined above which corresponding actions can be initiated, but below which corresponding actions may not be initiated. Such a threshold can, for example, be determined empirically, set by a user preference, or the like. As discussed in more detail below with reference to block 312, such a threshold can be dynamically adjusted based on a variety of perceived events and circumstances.

[0047] If the confidence level from block 306 does not exceed the threshold as determined at decision block 308 (e.g., the “no” branch), process 300 can return to block 302 and proceed by continuing to receive and sample audio. In some examples, process 300 can loop continuously to always listen for a speech trigger. In other examples, process 300 can loop to listen for a speech trigger during certain times (e.g., typical awake hours, work periods, social times, etc.). In still other examples, process 300 can loop to listen for a speech trigger for a certain non-zero time period following an event (e.g., power on, screen wake, alarm sounding, notification receipt, etc.). In yet other examples, process 300 can loop to listen for a speech trigger based on a variety of other factors (e.g., while a device is connected to a power source, while a user is engaged with a device, while a screen is on, during a scheduled calendar meeting, outside of scheduled calendar meetings, etc.).

[0048] If the confidence level from block 306 does exceed the threshold as determined at decision block 308 (e.g., the “yes” branch), process 300 can proceed to block 310. At block 310, a virtual assistant can be triggered to receive a user command from the audio input. For example, with a confidence level above the threshold indicating that a speech trigger was likely uttered and received, a virtual assistant can be triggered to receive a user command from the audio input. In some examples, triggering the virtual assistant can include initiating a virtual assistant session with a user. A virtual assistant session can, for example, include a prompt (e.g., a played sound, displayed image, displayed text, illuminated indicator light, etc.) to notify the user that a trigger was recognized and the system is ready to receive a command. A user can then utter a command or request for the virtual assistant. As discussed above, user intent can then be determined based on the received user command or request, and the command associated with the determined user intent can be executed or a response to the request can be provided.

[0049] In other examples, a virtual assistant session can be initiated without a particular prompt, and the virtual assistant can receive a user command from the audio input immediately following a trigger phrase. For example, a user can utter a trigger phrase immediately followed by a command or question (e.g., “Hey Siri, tell my mom I'm on my way.”). With a sufficiently high confidence level that a trigger phrase was spoken, the virtual assistant can be triggered to receive the subsequent audio input as a command (e.g., “Tell my mom I'm on my way”) without first requiring a prompt or other delay. In some examples, however, a brief prompt, silent prompt (e.g., text display, illuminated light, etc.), or other unobtrusive prompt can be issued quickly in between the trigger phrase and the command or question. In still other examples, as noted above, trigger phrases themselves can include commands or questions, and the virtual assistant can be caused to receive the commands or questions from the audio input at block 310 (e.g., from a buffer or the like).

[0050] At block 312, the speech trigger confidence level threshold used at decision block 308 can be dynamically adjusted in response to perceived events. In some examples, the threshold can be lowered, which can increase the sensitivity of the trigger, thus increasing the likelihood that audio input will be recognized as a trigger. In other examples, the threshold can be raised, which can decrease the sensitivity of the trigger, thus decreasing the likelihood that audio input will be recognized as a trigger. The threshold can be dynamically raised and/or lowered in response to a variety of perceived events, conditions, situations, and the like. For example, the threshold can be lowered to increase the likelihood of triggering when perceived events suggest that a user is likely to utter a speech trigger, a false trigger would be minimally inconvenient (e.g., while in a noisy environment), a missed trigger would be relatively inconvenient (e.g., while driving), or the like. Conversely, the threshold can be raised to decrease the likelihood of triggering when perceived events suggest that a user is unlikely to utter a speech trigger, a false trigger would be especially undesirable (e.g., during an important meeting or in a quiet environment), a missed trigger would be minimally inconvenient, or the like.

[0051] In one example, the threshold can be lowered when a missed trigger appears likely and a retry is anticipated. For example, a user may utter a trigger phrase (e.g., “Hey Siri”), but the confidence level determined at block 306 of process 300 may be below the threshold and insufficient to trigger the virtual assistant (e.g., due to soft speaking, interference, distance from a device, etc.). A user may be likely to retry and speak the trigger phrase a second time (e.g., more clearly, louder, closer to a device, etc.), and it can be desirable to ensure that the repeated trigger is not missed again. The confidence level determined during the missed trigger can be within a predetermined range of the threshold without exceeding the threshold, such as a few points below or within a certain percentage of the threshold. For instance, for a threshold of 70, a predetermined range can be defined as 50-69. In this example, the missed trigger may have been scored at a 65, which would be insufficient to trigger the virtual assistant. Given the perceived trigger within the predetermined range of 50-69, however, the threshold can be lowered such that the repeated trigger phrase can be more likely to trigger the virtual assistant the second time the user utters the phrase. The threshold can thus be lowered based on a prior sampling of the audio input having a confidence level within a predetermined range of the threshold without exceeding the threshold. It should be appreciated that the predetermined range can be determined empirically, set by a user, or the like.

[0052] In another example, the threshold can be lowered or raised based on detecting content shown on a display. For example, the threshold can be lowered when a screen is activated or awake, when certain content is displayed that is common subject matter for speech requests, when a notification is received or displayed, when a user is accessing virtual assistant services, or the like. The relevant display can include touchscreen 246 of user device 102 or any other display (e.g., a television display, laptop display, control panel display, etc.). In some examples, the threshold can be raised or lowered based on the type of content shown or similarly based on a user's interactions with on-screen elements (e.g., selecting music to be played, reading emails, responding to messages, etc.).

[0053] FIG. 4A illustrates user device 102 with touchscreen 246 showing virtual assistant interface 422 that can be used in adjusting a speech trigger threshold. In one example, the threshold can be lowered whenever interface 422 is displayed or whenever a user is otherwise interacting with a virtual assistant or virtual assistant services (e.g., when any content associated with a virtual assistant is displayed). In particular, as virtual assistant interactions can extend beyond one user query, the likelihood that additional queries will be received during the same interaction can be used to lower the speech trigger threshold, thereby increasing the likelihood of triggering for subsequent queries. Interface 422 can include, for example, assistant greeting 424 prompting the user for a command, transcribed user speech 426 including a user query (e.g., “Will it rain today?”), and assistant response 428 responding to the query (e.g., “There is a 60% chance of rain in the forecast.”). While interface 422 is displayed, the threshold for triggering the virtual assistant to again receive a command can be lowered to readily detect follow-on interactions. For example, while interface 422 is displayed, a user can utter “Launch Photos,” which can be recognized as a trigger phrase in some examples. The lowered threshold can make it more likely that the utterance will trigger and cause the virtual assistant to receive the command and execute the associated user intent.

[0054] In another example, the threshold can be lowered based on whether the virtual assistant is active. In some examples, whether the virtual assistant is active can be apparent from a display (e.g., such as interface 422 on touchscreen 246). In other examples, however, the virtual assistant can be active without information being displayed (e.g., active on a device without a display, operating in the background, operating in a vehicle without displaying information, etc.). As with interface 422 of FIG. 4A, detecting that the virtual assistant is active can be used to lower the threshold to increase in the likelihood of triggering on received trigger utterances. This can increase the likelihood that follow-on utterances will correctly trigger the virtual assistant to receive a subsequent query.

[0055] FIG. 4B illustrates user device 102 with touchscreen 246 showing exemplary display content that can be used in adjusting the speech trigger threshold. In one example, the speech trigger threshold can be adjusted in response to detecting that a notification has been received, issued, displayed, sounded, or the like. For example, notification interface 432 can be displayed (e.g., animated to slide down from the top of touchscreen 246 as indicated by the arrow) showing that new email 434 was received. In response to detecting display of notification interface 432, receipt of new email 434, sounding of a notification, or another indication that a notification has been provided to the user, the speech trigger threshold can be lowered. In some instances, a notification can prompt a user to issue related virtual assistant commands, such as reading out an email, displaying a new message, responding to a received message, answering an incoming call, rejecting an incoming call, or the like. The speech trigger threshold can be lowered upon detecting a notification such that triggering the virtual assistant can be more likely given the increased likelihood that a user would desire assistance immediately following a received notification.

[0056] Any type of notification can be detected to cause the speech trigger threshold to be lowered, including incoming calls, new messages, application notifications, alerts, calendar reminders, or the like. In addition, in some examples, different types of notifications can cause the threshold to be lowered by different amounts. For example, the threshold can be lowered more for an incoming call than for an application notification because a user may be more likely to issue voice commands associated with an incoming call than for an application notification. In other examples, the threshold can be raised in response to detecting a notification. For example, the threshold can be raised in response to a calendar reminder that a meeting is starting such that the virtual assistant can be less likely to falsely trigger and interrupt the meeting. Notifications can thus be used in a variety of ways to adjust the speech trigger threshold.

[0057] In some examples, notifications can be used as pseudo triggers to allow users to interact with a virtual assistant without first uttering a speech trigger. For example, in response to receiving a notification (e.g., new email 434 shown in notification interface 432), the virtual assistant can be triggered to receive a user command from the audio input without detecting receipt of a spoken trigger. Should a user utter a command following the notification (e.g., “read that to me”), the user's intent can be determined and the command can be executed (e.g., the newly received email can be read out) without first requiring a particular speech trigger. Should a user not utter a command within a certain time, a user intent is unlikely to be determined from the input, and no virtual assistant action may take place. The time to wait for a spoken command can vary based on the notification, such as listening for commands for a longer duration when multiple notifications are received or notifications are complex (e.g., calendar invitations, messages, etc.) than when a single, simple notification is received (e.g., a reminder, alert, etc.). In some examples, the virtual assistant can thus be triggered and listen for commands in response to receiving a notification without first requiring a particular speech trigger.

[0058] In other examples, the speech trigger threshold can be adjusted in other ways based on different types of display content. FIG. 4B, for example, illustrates music interface 436 with music options 438 on touchscreen 246 of user device 102. In one example, the speech trigger threshold can be lowered based on music interface 436 being displayed or while music interface 436 is displayed. A virtual assistant can perform various operations on behalf of a user, including playing particular music. Thus, while music interface 436 is displayed, it may be more likely that a user may utter a speech trigger and request playback of one of music options 438 or other music. By lowering the threshold based on the display of music interface 436, triggering the virtual assistant can be more likely, thus reducing the risk of missed triggers when a spoken trigger is likely.

[0059] Various other types of displayed content can similarly be used to raise or lower the speech trigger threshold. For example, displayed content that may be associated with common virtual assistant requests can cause the speech trigger threshold to be lowered, while displayed content that may be unassociated with common virtual assistant requests can cause the speech trigger threshold to be raised. The speech trigger threshold can be lowered, for example, based on display of email messages, text messages, music lists, application lists, calendar entries, contacts, a phone application, stocks, weather information, social media feeds, restaurants, movies, maps, addresses, task lists, reminders, notes, settings, certain applications, or the like. The speech trigger threshold can be raised, for example, based on display of an electronic book, a playing movie, some games, certain applications, or the like. The amount by which the speech trigger threshold is raised or lowered can also vary based on the type of displayed content, such as lowering the threshold more for display of a phone application than for display of a list of stocks. In some examples, the amount of raising or lowering can be associated with how common virtual assistant queries are that are associated with the particular displayed content. The speech trigger threshold can thus be adjusted in a variety of ways based on a variety of displayed content.

[0060] In another example, the speech trigger threshold can be adjusted based on various sensors associated with various devices. For example, the speech trigger threshold can be raised or lowered based on detecting a user's proximity to an electronic device (e.g., using proximity sensor 214 and/or camera 222 of user device 102). When, for instance, no user is detected to be present, the threshold can be raised to avoid triggering on spurious noises or sounds. On the other hand, when a user is detected to be near a device, the threshold can be lowered to increase the likelihood of triggering on an uttered speech trigger. In some examples, a proximity sensor can be used to detect the presence or absence of a user. In other examples, a camera can be used to detect the presence or absence of a user. In still other examples, facial recognition can be performed to recognize a particular user in a camera view, and the speech trigger threshold can be adjusted based on the presence or absence of a user associated with a device (e.g., the device owner, an authorized user, a relative, etc.). User proximity can be determined in a variety of locations by a variety of devices, including by a device in a vehicle, a device in a hallway, a device in a room at a house or business, a device in an outdoor area, or the like.

[0061] In one example, a sound profile can be analyzed to determine user presence and adjust a speech trigger threshold accordingly. For example, audio received at a device can be analyzed to determine the types of sounds making up the sound profile. Based on the sound profile, a user's proximity can be determined. Sounds associated with human presence, for instance, can be indicative of nearby users. Such sounds can include human speech, breathing, movement (e.g., walking), cleaning, cooking, or the like. In contrast, other sounds can be associated with the absence of human presence, such as white noise, fan noise, noise from electronics, animal sounds, echoes, or the like. In one example, whether a user is near a device can be determined based on the detected sound profile containing sounds associated with human presence as opposed to noises associated with the absence of a person. Based on the determined proximity of a user, a speech trigger threshold can be lowered to more readily trigger on uttered speech triggers. In contrast, based on the determined absence of a user, a speech trigger threshold can be raised to avoid triggering on spurious sounds when a user may be unlikely to be present.

[0062] Sensor data can be collected from multiple different devices and used alone or in combination to adjust the speech trigger threshold. In one example, a camera, motion sensor, or the like associated with a security system, gaming system, television system, or the like can be used to detect the presence or proximity of a user. The detected absence of a user can be used to raise the trigger threshold to avoid false triggers, while the detected presence of a user can be used to lower the trigger threshold to avoid missed triggers. In one example, a speech-enabled device can use one or more sensors to detect user presence or proximity to then adjust its own speech trigger threshold. For instance, a speech-enabled mobile device, speech-enabled television control system, speech-enabled gaming system, speech-enabled security system, or the like can include a motion detector, proximity sensor, camera, or other sensors. Based on the sensor data, the speech trigger threshold for the various systems or devices can be adjusted. When, for example, a user enters a room or the area near a device and is detected by one or more sensors, the speech trigger threshold can be lowered to avoid missed triggers from the present user. When the user leaves the room or area and the user's absence is apparent or presumed from sensor data, the speech trigger threshold can be raised to avoid false triggers from spurious noises or sounds. In this manner, various types of devices can use available sensor information to dynamically adjust a speech trigger threshold based on user proximity or presence.

[0063] In other examples, multiple devices that are nearby and in communication (e.g., via Bluetooth, Wi-Fi, or other communication channels) can share sensor data or trigger threshold information to adjust speech trigger thresholds for the various devices. FIG. 5 illustrates multi-device system 540, which can include interactions among the various devices to dynamically adjust a speech trigger threshold for one or more of the devices. In the illustrated example, a variety of devices and multiple users can be present, and speech trigger thresholds for various speech-enabled devices can be adjusted based on sensor information from the various devices. Multi-device system 540, for instance, can include multiple speech-enabled devices, such as TV set-top box 546, security system control panel 554, user device 102, and tablet computer 564. The various devices can be associated with a variety of different sensors and displays. TV set-top box 546, for example, can be connected to display 542 and camera 544, which can be used for facial recognition, gesture detection, presence detection, photography, filming, or the like. Security system control panel 554 can be associated with camera 556, which can likewise be used for facial recognition, gesture detection, presence detection, or the like. User device 102 and tablet computer 564 can include a variety of sensors that can likewise be used to detect user presence.

[0064] In one example, sensor data from one device can be used to adjust a speech trigger threshold of another device. For instance, TV set-top box 546 with camera 544 or security system control panel 554 with camera 556 can detect the presence of user 550. The presence of user 550 (or the presence of an unidentified person) can then be communicated to other nearby devices, such as user device 102 and tablet computer 564. Based on the detected presence of user 550, user device 102 and/or tablet computer 564 can lower their speech trigger thresholds to make triggering more likely given a user is present, thus avoiding missed triggers. In another example, the absence of any user detected by security system control panel 554 with camera 556 can be communicated to TV set-top box 546, which can raise its speech trigger threshold to avoid false triggers given the detected absence of a user. In yet another example, the presence of user 558 can be detected by user device 102 or tablet computer 564 and communicated to one or both of TV set-top box 546 and security system control panel 554, which can in turn lower their speech trigger thresholds to avoid missed triggers given a user's presence is detected. Sensor data from various devices can thus be communicated to various other devices and used to adjust a speech trigger threshold.

[0065] In another example, distances between users and multiple devices can be used to determine which device should trigger upon detection of a trigger phrase. For example, in multi-device system 540, multiple devices may be able to detect a speech trigger from either user 550 or user 558. To avoid multiple speech-enabled devices triggering from the same utterance, relative user proximity to various devices can be determined and used to ensure that the nearest device is the most likely to trigger. In one example, devices can communicate relative distances to determine which device is nearest a user. TV set-top box 546 can, for example, determine approximate distance 548 using camera 544 (e.g., based on image size of user 550, time of flight ranging, proximity sensor values, etc.). Security system control panel 554 can determine approximate distance 552 using camera 556 (e.g., based on image size of user 550, time of flight ranging, proximity sensor values, etc.). The two devices can compare the determined distances 548 and 552 to determine that user 550 may be nearer security system control panel 554 given that distance 552 is less than distance 548. Based on the user's proximity to security system control panel 554 and relative distance from TV set-top box 546, a speech trigger threshold associated with security system control panel 554 can be lowered to make it more likely that it will trigger in response to a detected speech trigger phrase. In some examples, a speech trigger threshold associated with TV set-top box 546 can be raised to decrease the likelihood that it will trigger in response to a detected speech trigger phrase.

[0066] In other examples, distances 548 and 552 can be approximated in other ways to determine which device should trigger in response to detecting a speech trigger. For example, upon detecting a speech trigger phrase, TV set-top box 546 and security system control panel 554 can compare the confidence values determined from the detected speech trigger (e.g., at block 308 of process 300). The device having the higher confidence value can be selected to trigger in response to the speech trigger. In another example, volume levels of audio input received by the two devices can be compared, and distances 548 and 552 can be approximated from the relative volume levels to determine which of the two devices should respond to a speech trigger. This can be done, in some examples, by lowering the trigger threshold for the nearer device and raising the trigger threshold for the more distant device. In the example of FIG. 5, any of a variety of sensor data (e.g., from cameras, microphones, proximity sensors, motion detectors, etc.) can be used and compared between TV set-top box 546 and security system control panel 554 to approximate distances 548 and 552, or to determine whether user 550 is likely closer to one device or the other. To cause the appropriate device to respond to a speech trigger, the speech trigger threshold can be lowered for the device determined to be nearest the user (e.g., security system control panel 554 at distance 552), while the speech trigger threshold can be raised for devices determined to be more distant (e.g., TV set-top box 546 at distance 548). Distances 548 and 552 can also be approximated in any of a variety of other ways, including time of flight ranging, echo location, signal strength determination, proximity sensor values, user sizes in camera views, and the like.

[0067] In another example, relative distances can be determined for mobile or handheld devices and used to adjust a speech trigger threshold of those devices. For example, user 558 can utter a trigger phrase near user device 102 and tablet computer 564 as illustrated in FIG. 5. To avoid both devices triggering in response to a speech trigger, the proximity of user 558 to the devices can be compared, and the speech trigger threshold of the nearest device can be lowered while the speech trigger threshold of the more distant device can be raised. In the illustrated example, distances 560 and 562 can be determined in any of a variety of ways. For example, approximate distances 560 and 562 can be determined based on time of flight ranging, echo location, signal strength determination, proximity sensor values, user sizes in camera views, volume levels of speech input, temperature, motion, or any of a variety of other sensor data. The approximated distances 560 and 562 can be shared between user device 102 and tablet computer 564. Based on the proximity of user 558 to user device 102 at distance 560 compared to distance 562 to tablet computer 564, the speech trigger threshold of user device 102 can be lowered while the speech trigger threshold of tablet computer 564 can be raised. By adjusting the thresholds of the devices, speech triggers from user 558 can be more likely to trigger on user device 102 than on tablet computer 564, thus avoiding triggering on multiple devices from the same utterance.

[0068] In other examples, multiple devices of various types can communicate sensor information to ensure that a speech-enabled device near a user triggers while other nearby speech-enabled devices do not. For example, in multi-device system 540, any of the various devices could trigger in response to speech triggers uttered by either user 550 or user 558. As such, the various distances can be determined and shared among the devices, and it can be determined which devices are near users and which are more distant from users. In one example, speech trigger thresholds of the various speech-enabled devices can be adjusted according to the shared distances (e.g., a speech trigger threshold of user device 102 can be lowered, a speech trigger threshold of security system control panel 554 can be lowered, a speech trigger threshold of tablet computer 564 can be raised, and a speech trigger threshold of TV set-top box 546 can be raised). In this manner, devices determined to be near users can be more likely to trigger than devices that are determined to be more distant from users, even though multiple devices near a user could otherwise trigger from the same utterance.

[0069] In still other examples, which of multiple devices triggers in response to detecting a trigger phrase can be determined based on comparing confidence levels among the various devices, comparing volume levels of speech received at various devices, comparing various other types of sensor data among the various devices, or the like. For example, if two or more devices could trigger based on a single utterance, before triggering, the devices could share sensor data to determine which of the devices should trigger and which should not.

[0070] In yet other examples, various other sensors and user data can be used in determining whether or not to trigger in response to a particular utterance. For example, facial recognition can be used to determine if a user's face is directed toward TV set-top box 546 rather than toward security system control panel 554. Based on the facial recognition, a speech trigger threshold of TV set-top box 546 can be lowered and/or a speech trigger threshold of security system control panel 554 can be raised. In another example, based on the facial recognition, TV set-top box 546 can trigger instead of security system control panel 554 in response to a detected speech trigger that may have been detected by both devices (e.g., by communicating trigger detection between devices before triggering).

[0071] It should thus be appreciated that multiple devices can communicate sensor information, speech trigger detection, confidence levels, and the like, and such devices can coordinate responding to detected speech triggers and/or can adjust speech trigger thresholds to favor some devices over others. It should further be understood that multi-device system 540 and the examples discussed illustrate possible interactions among multiple devices, but many other devices with various other interactions can likewise employ the techniques discussed herein for adjusting speech trigger thresholds and/or favoring triggering at one device over another.

[0072] In another example, a speech trigger threshold can be adjusted based on detected motion (e.g., from an accelerometer and/or gyroscope of user device 102). In some examples, a user may be more likely to utter a speech trigger after moving a device (e.g., lifting a device out of a bag, pulling a device out of a pocket, picking a device up off of a table, etc.). Based on the detected motion, a speech trigger threshold can be lowered to more readily trigger on uttered speech triggers. In other examples, a user may be less likely to utter a speech trigger when a device is in regular or somewhat constant motion (e.g., while a user is walking, when a user is exercising, etc.). Based on those detected motions, a speech trigger threshold can be raised to decrease the likelihood of triggering.

[0073] In one example, a speech trigger threshold can be adjusted based on a time of day or a scheduled calendar event. For example, a speech trigger threshold can be lowered during non-work hours when a user may be more likely to interact with a device using speech. In contrast, a speech trigger threshold can be raised during working hours when a user may be less likely to interact with a device using speech. In some examples, a user can set a preference for sensitivity during certain hours or situations. In other examples, a speech trigger threshold can be adjusted based on calendar events or recognized situations. For example, a speech trigger threshold can be raised during a work meeting, class, date, or the like to avoid false triggers that may interrupt. In another example, a speech trigger threshold can be lowered during an outing with friends, a commute, certain leisure activities, or the like as speech interactions may be more likely in those situations.

[0074] In another example, a speech trigger threshold can be retroactively adjusted based on processing of sampled audio input. For example, sampled audio input can be analyzed to determine user speech and/or user intent. If an actionable command is detected in the sampled audio input, a speech trigger threshold can be retroactively lowered such that the actionable command can be received by a virtual assistant and executed. For instance, should a user utter a trigger phrase immediately followed by an actionable command (e.g., “Hey Siri, call dad”), the speech trigger may be missed for any of a variety of reasons, but the actionable command can be identified in received audio, the speech trigger threshold can be retroactively lowered, and the actionable command can be received and executed by the virtual assistant as a result.

[0075] In other examples, a speech trigger threshold can be adjusted in response to various user actions. For example, after a false trigger, a user can utter a phrase to indicate the trigger was undesired (e.g., “not now,” “false trigger,” “mistake,” “quiet,” “don't interrupt,” “turn off,” etc.). The speech trigger threshold can then be raised to avoid further false triggers. In another example, the speech trigger threshold can be raised based on the position or location of the device. For example, proximity of an object to an electronic device can be detected, such as when an electronic device is face down on a table, in a pocket, in a purse, against a user's face, in a sleeve, against a case, or the like. Based on the proximity of an object to the device, the speech trigger threshold can be raised given speech interactions may be less likely or undesired based on the positioning of the device. In another example, a speech trigger threshold can be raised based on particular detected device motions, such as a user flipping a device onto its face on a table, a user shaking a device in a particular way, a user sliding a device into a pocket, or the like.

[0076] In another example, a speech trigger threshold can be adjusted in response to detected device settings. For example, a speech trigger threshold can be raised in response to detecting a vibrate, silent, and/or do not disturb setting associated with an electronic device (e.g., a setting on user device 102 to vibrate only, keep silent, and/or avoid all notifications, including visual). In some examples, threshold adjustments can differ based on device setting, such as raising the speech trigger threshold a first amount based on a vibrate setting, and raising the speech trigger threshold a second larger amount based on a do not disturb setting. In this manner, false triggers that may undesirably interrupt can be avoided based on cues from a user selecting particular device settings.

[0077] In any of the various examples discussed herein, adjustments to a speech trigger threshold can be temporary or permanent. In some examples, a speech trigger threshold can be restored to a predetermined level after a predetermined time (a time greater than zero) has elapsed. For example, a threshold adjustment can be maintained for a predetermined non-zero time, which can be set for all adjustments or can be different for different events that caused the threshold to be adjusted. A predetermined time to hold an adjustment can be longer, for example, when a user actively indicates an importance of avoiding false triggers (e.g., based on a do not disturb setting, a user uttering “not now,” an important meeting, etc.). A predetermined time to hold an adjustment can be shorter, for example, when an event that caused the adjustment is likely temporary (e.g., a potential missed trigger, a notification, displayed content, etc.). Threshold adjustments can thus be held for non-zero times of varying durations, after which the threshold can be automatically restored to a predetermined level.

[0078] In other examples, a speech trigger threshold can be restored to a predetermined level after the end of an event. For example, a speech trigger threshold can be restored to a predetermined level at the end of a calendar meeting, at the end of a workday, at the end of a leisure activity, at the end of a commute, upon removal of a setting, upon a change in location, upon movement of a device, or the like. A device can detect the ends of such events in a variety of ways, such as based on calendar settings, device movements, user interactions, sound profiles, or the like.

[0079] In some examples, the predetermined level of a speech trigger threshold can be adjusted over time based on usage. For example, the set or restore point of the threshold can be adjusted over time based on interactions with a user, missed triggers, false triggers, or the like. In one example, based on user feedback of frequent false triggers (e.g., a user silencing a device, uttering “not now,” identifying a false trigger with a button, not following a prompt with a request, etc.), the predetermined level of a speech trigger can be raised over time to make triggering less likely, and the level can continue to be raised based on how frequent false triggers are going forward. In contrast, based on user feedback of frequent missed triggers (e.g., repeated trigger phrases, manual assistant activation, manual indications, etc.), the predetermined level of a speech trigger can be lowered over time to make triggering more likely, and the level can continue to be lowered based on how frequent missed triggers are going forward. In another example, the predetermined level can be lowered over time for users who interact frequently using speech triggers and can be raised over time for users who interact infrequently using speech triggers. In other examples, users can manually set the predetermined level for speech triggers (e.g., a sensitivity for speech triggers).

[0080] In any of the various examples herein, the amount by which a speech trigger threshold is raised or lowered can vary according to the perceived event causing the adjustment or a variety of other factors. For example, the adjustment amount can be set to have an optimal impact according to the perceived event warranting the adjustment (e.g., varying the adjustment amount with the severity, duration, impact, or the like of the perceived event). In another example, the adjustment amount can be set empirically based on effectiveness in minimizing missed and false triggers. In still other examples, the adjustment amount can be set by user preference.

[0081] It should be understood that a speech trigger threshold can be raised or lowered in response to any of the perceived events discussed herein. It should be further understood that a speech trigger threshold can be adjusted based on multiple perceived events (e.g., lowered by summed amounts based on concurrent events, adjusted based on offsetting events, etc.). In this manner, a speech trigger threshold can be dynamically adjusted to, for example, reduce missed and false triggers and provide an enjoyable and efficient user experience.

[0082] In addition, in any of the various examples discussed herein, various aspects can be personalized for a particular user. The various processes discussed herein can be modified according to user preferences, contacts, text, usage history, profile data, demographics, or the like. In addition, such preferences and settings can be updated over time based on user interactions (e.g., frequently uttered commands, frequently selected applications, etc.). Gathering and use of user data that is available from various sources can be used to improve the delivery to users of invitational content or any other content that may be of interest to them. The present disclosure contemplates that in some instances, this gathered data can include personal information data that uniquely identifies or can be used to contact or locate a specific person. Such personal information data can include demographic data, location-based data, telephone numbers, email addresses, home addresses, or any other identifying information.

[0083] The present disclosure recognizes that the use of such personal information data, in the present technology, can be used to the benefit of users. For example, the personal information data can be used to deliver targeted content that is of greater interest to the user. Accordingly, use of such personal information data enables calculated control of the delivered content. Further, other uses for personal information data that benefit the user are also contemplated by the present disclosure.

[0084] The present disclosure further contemplates that the entities responsible for the collection, analysis, disclosure, transfer, storage, or other use of such personal information data will comply with well-established privacy policies and/or privacy practices. In particular, such entities should implement and consistently use privacy policies and practices that are generally recognized as meeting or exceeding industry or governmental requirements for maintaining personal information data as private and secure. For example, personal information from users should be collected for legitimate and reasonable uses of the entity and not shared or sold outside of those legitimate uses. Further, such collection should occur only after receiving the informed consent of the users. Additionally, such entities would take any needed steps for safeguarding and securing access to such personal information data and ensuring that others with access to the personal information data adhere to their privacy policies and procedures. Further, such entities can subject themselves to evaluation by third parties to certify their adherence to widely accepted privacy policies and practices.

[0085] Despite the foregoing, the present disclosure also contemplates examples in which users selectively block the use of, or access to, personal information data. That is, the present disclosure contemplates that hardware and/or software elements can be provided to prevent or block access to such personal information data. For example, in the case of advertisement delivery services, the present technology can be configured to allow users to select to “opt in” or “opt out” of participation in the collection of personal information data during registration for services. In another example, users can select not to provide location information for targeted content delivery services. In yet another example, users can select not to provide precise location information, but permit the transfer of location zone information.

[0086] Therefore, although the present disclosure broadly covers use of personal information data to implement one or more various disclosed examples, the present disclosure also contemplates that the various examples can also be implemented without the need for accessing such personal information data. That is, the various examples of the present technology are not rendered inoperable due to the lack of all or a portion of such personal information data. For example, content can be selected and delivered to users by inferring preferences based on non-personal information data or a bare minimum amount of personal information, such as the content being requested by the device associated with a user, other non-personal information available to the content delivery services, or publicly available information.

[0087] In accordance with some examples, FIG. 6 shows a functional block diagram of an electronic device 600 configured in accordance with the principles of the various described examples. The functional blocks of the device can be implemented by hardware, software, or a combination of hardware and software to carry out the principles of the various described examples. It is understood by persons of skill in the art that the functional blocks described in FIG. 6 can be combined or separated into sub-blocks to implement the principles of the various described examples. Therefore, the description herein optionally supports any possible combination or separation or further definition of the functional blocks described herein.

[0088] As shown in FIG. 6, electronic device 600 can include an input interface unit 602 configured to receive information (e.g., a microphone to record audio). Electronic device 600 can further include an output interface unit 604 configured to output information (e.g., a display to display content). Electronic device 600 can further include processing unit 606 coupled to input interface unit 602 and output interface unit 604. In some examples, processing unit 606 can include an audio receiving unit 608, an audio sampling unit 610, a confidence level determining unit 612, a virtual assistant triggering unit 614, and a threshold adjustment unit 616.

[0089] Processing unit 606 can be configured to dynamically adjust a speech trigger threshold. Processing unit 606 can be configured to receive audio input via a microphone (e.g., through input interface unit 601 using audio receiving unit 608). Processing unit 606 can be further configured to sample the received audio input (e.g., using audio sampling unit 610). Processing unit 606 can be further configured to determine (e.g., using confidence level determining unit 612) a confidence level that the sampled audio input comprises a portion of a spoken trigger. Processing unit 606 can be further configured to, in response to the confidence level exceeding a threshold, trigger a virtual assistant (e.g., using virtual assistant triggering unit 614) to receive a user command from the audio input. Processing unit can be further configured to dynamically adjust the threshold (e.g., using threshold adjustment unit 616) in response to a perceived event.

[0090] In some examples, triggering the virtual assistant comprises initiating a virtual assistant session with a user. In other examples, processing unit 606 can be further configured to determine a user intent based on the received user command and execute a command associated with the determined user intent. In some examples, processing unit 606 can be further configured to lower the threshold in response to the perceived event. In one example, the perceived event comprises determining, from a prior sampling of the audio input, a confidence level that the prior sampled audio input comprises a portion of the spoken trigger; wherein the confidence level for the prior sampled audio input is within a predetermined range of the threshold without exceeding the threshold. In another example, the perceived event comprises detecting content shown on a display. In yet another example, the perceived event comprises detecting display of content associated with the virtual assistant. In other examples, the perceived event comprises detecting that the virtual assistant is active.

[0091] In some examples, the perceived event comprises detecting a user's proximity. In one, the user's proximity is detected based on one or both of a camera or proximity sensor associated with the electronic device. In another example, the user's proximity is detected based on one or more of a camera or proximity sensor associated with a device other than the electronic device. In another example, the user's proximity is detected based on a motion detector. In another example, the user's proximity is detected based on a sound profile comprising sounds associated with human presence. In another example, the user's proximity is detected at the electronic device and at least one other device, and the threshold is lowered at the electronic device based on closer user proximity to the electronic device than the at least one other device. In yet another example, the user's proximity is detected in a vehicle. In other examples, the perceived event comprises detected motion. In still other examples, the perceived event comprises receiving a notification. In one example, the perceived event comprises a time of day or a scheduled calendar event. In another example, the perceived event comprises detecting an actionable user command in the sampled audio input.

[0092] In some examples, processing unit 606 can be further configured to raise the threshold in response to the perceived event. In one example, the perceived event comprises detecting a user command to dismiss the virtual assistant. In another example, the perceived event comprises detecting proximity of an object to the electronic device. In another example, the perceived event comprises a scheduled meeting. In another example, the perceived event comprises detecting absence of a user. In yet another example, the perceived event comprises detecting a do not disturb setting associated with the electronic device.

[0093] In some examples, the threshold is restored to a predetermined level after a predetermined time has elapsed. In one example, the predetermined level is adjusted over time based on usage. In another example, processing unit 606 can be further configured to, in response to receiving a notification, trigger the virtual assistant to receive a user command from the audio input. In yet another example, the virtual assistant is triggered without detecting receipt of the spoken trigger.

[0094] Although examples have been fully described with reference to the accompanying drawings, it is to be noted that various changes and modifications will become apparent to those skilled in the art (e.g., modifying any of the systems or processes discussed herein according to the concepts described in relation to any other system or process discussed herein). Such changes and modifications are to be understood as being included within the scope of the various examples as defined by the appended claims.