Background
1. Field of the Invention
The present invention relates to telephone answering devices and sound signal manipulation, and more particularly to the means by which a voice signal may be better presented to an analog/digital converter in a digital telephone answering device.
2. Description of the Prior Art
Telephone answering devices (TADs) are well known and in common use throughout the United States. Recently, advancement in digital technology has influenced recent TAD design and production. Instead of using cassette tapes to record outgoing and incoming messages, TADs are now available that use digital memory storage as a recording medium. Digital memory storage, namely random access memory (RAM), requires the conversion of an analog voice signal generated by a microphone or phone line into a digital representation. This is not difficult, but does require a good clear signal for optimum performance.
Such good clear signals are easily generated by microphones attached to digital telephone answering devices (DTADs), but the same is not true for signals coming into the DTAD from the phone line. On present phone lines, voice signals from telephone lines have a dynamic range of about thirty or more decibels (dB). For this reason, some calls can be very clear while others can be faint or have much noise.
Standard analog to digital converters are eight bits "wide" and have a dynamic range of forty-eight decibels. Between the dynamic ranges of the phone line and the analog to digital converter is a difference of eighteen decibels. Thus, under worst case conditions, the effective dynamic range of the ADC would be limited to only eighteen dB. This would have severe voice quality consequences that make an eight bit ADC unusable. Conversely, if the dynamic range of the signal from the phone line was reduced to three dB, the ADC would have forty-five dB with which to work and eight bits would easily handle voice signals from the phone line.
Beyond the limitations of current eight bit ADCs with respect to phone line dynamic range, the use of sophisticated speech compression algorithms require clear signals for their operation. Without such clear signals, the use of speech compression in the storing and retrieving the digitized analog signal results in an overall degradation of the voice signal. The signal degradation is not experienced by clear signals, but only those that suffer a significant loss of integrity upon digitization, such as low level signals.
When an eight bit ADC attempts to convert a phone line signal from analog to digital representation, the signals that are close to the ADCs input range are fairly well converted. When the whole of the signals transmitted to a DTAD over the phone line are of low intensity, the digital signal created by the ADC will lose some of the information contained in the analog signal. As the analog signal from the phone line is of low intensity, only a small portion of the ADC's digitizing range is used. This leads to what is called "quantization noise" in the digitized signal and is noticeable when the digital signal is converted back into an analog signal upon playback. The quantization noise arises from the resolution available in an eight bit ADC. When the signal is low, the audio signal resolving power of an eight bit ADC is limited. In an extreme example of a faint signal, only four or five of the eight bit ADC's two hundred and fifty-six quantizing levels might be used for the whole of the signal.
When an analog signal is sampled by the ADC, the ADC changes the analog signal into the closest digital representation available, even though there is some disparity between the analog signal and its closest available digital representation. The more bits available to the ADC, the closer the digital representation will be to the analog signal. The loss of part of the analog signal experienced in digitization is called the quantization noise and can be modeled as the introduction of an unwanted signal when the analog signal is quantified by digitization. When the quantization noise of a digitized analog signal is high, the resulting analog signal created when the digital signal is played back has a tendency to become "granular" and reveal the discrete levels that were used to digitally record the original analog signal. When an analog signal dwells within a short range of digital levels, the quantization noise can be severe. Analog signals of low intensity with respect to the input range of the ADC have the tendency to dwell within such a short range.
One method that can be used to combat quantization noise is to increase the resolution of the ADC. By increasing the number of bits, an ADC has more levels available with which an analog signal can be represented. Where the number of bits can be increased, this is a successful means for reducing quantization noise. As a result, most DTADs demand that a twelve of thirteen bit ADC be used. The use of larger ADC's having more bits demands greater cost and more sensitive circuitry compared with eight bit systems. Additionally, many single chip microcontrollers used in DTADs have built-in ADCs and these ADC's are usually limited to only eight bits. If an eight bit ADC could be used, circuitry would be eliminated and costs could be reduced.
Another method is merely to use a fixed gain amplifier. In Millet, U.S. Pat. No. 4,794,638, issued on Dec. 27, 1988, a fixed gain amplifier is used. As the gain is fixed, no increase in the effective dynamic range is realized. High and low intensity signals are amplified by the same amount. Upon digitization, no improvement is seen in the loss of low intensity signals and quantization noise is not diminished.
In light of the decreased expense and design considerations, there is a need for a method by which eight bit ADC's may be used in DTADs with limited quantization noise for low intensity signals.
Summary of the Invention
The present invention lies in the use of an automatic gain controller in the form of a voltage controlled amplifier coupled to an averaging circuit that feeds back into the voltage controlled amplifier. The averaging circuit supplies the controlling voltage that keeps the amplifier output voltage tending towards a certain reference level. When the output of the amplifier is generally high, the output of the averaging circuit is also generally high and the reverse is also true for low signals. The averaging circuit output is fed back to the amplifier. For high averaging circuit output voltages corresponding to signals of high intensity, amplification of the signal by the amplifier is diminished or subdued. For low averaging circuit output voltages corresponding to low intensity signals, amplification of the signal by the amplifier is enhanced or augmented. The result is to boost the signal of low intensity signals significantly, compressing the dynamic range of the incoming phone signal. This reduces the variance of an incoming phone line signal for the analog to digital converter (ADC) present in a digital telephone answering device (DTAD). With reduced variance at its input signal, the effective dynamic range of the ADC is closer to its optimum dynamic range. For coupling with an ADC, a scaling circuit intermediates the signal from the automatic gain controller to the ADC.
Objects of the Invention
It is an object of the present invention to provide means for expanding the effective dynamic range of an analog to digital converter used in a digital telephone answering device.
It is another object of the present invention to provide such means that are cost effective.
It is still another object of the present invention to provide such means that allow the use of an eight bit analog to digital converter.
It is yet another object of the present invention to provide such means that are simple and easy to use.
These and other objects and advantages of the present invention will be apparent from a review of the following specification and accompanying drawings.
Brief Description of the Drawings
FIG. 1 is a block diagram of the present invention showing the components and their relative positions.
Detailed Description of the Preferred Embodiments
As shown in FIG. 1, an analog to digital converter (ADC) 10 of a digital telephone answering device (DTAD) is coupled by its input to a phone line 12. The phone line 12 is connected to the input of an automatic gain controller (AGC) 14. The output of the AGC 14 is connected to the input of a scaling circuit 16. The output of the scaling circuit 16 is connected to the input of the ADC 10.
The AGC 14 is comprised of two components: a voltage controlled amplifier 18 and an averaging circuit 20. The voltage controlled amplifier 18 is connected by its input to the phone line 12. The output of the voltage controlled amplifier 18 is connected to the input of the averaging circuit 20 and the scaling circuit 16. The output of the averaging circuit 20 is connected to the voltage controlled amplifier 18 and serves as the voltage controlling the voltage controlled amplifier 18. Both voltage controlled amplifiers and averaging circuits are well known in the art.
The voltage controlled amplifier 18 amplifies the output signal of the phone line 12 according to the voltage applied on voltage controlled amplifier 18 by the averaging circuit 20. The voltage applied by the averaging circuit 20 is directly dependent upon the output of the voltage controlled amplifier 18. In this way, the voltage controlled amplifier 18 controls itself by feedback through the averaging circuit 20.
The averaging circuit 20 is present to prevent instantaneous response by the voltage controlled amplifier 18 to its own signal. With the averaging circuit 20, the long term average level of the phone line 12 signal is amplified rather than individual signals. The averaging circuit 20 feeds back to the voltage controlled amplifier 18 a signal representative of the voltage controlled amplifier 18 output for the previous two to four seconds.
The voltage controlled amplifier 18 amplifies the phone line 12 signal most when a low voltage is applied by the averaging circuit 20 upon the voltage controlled amplifier 18. A low voltage is applied by the averaging circuit 20 when the output signal of the voltage controlled amplifier 18 is low. This will cause the gain of voltage controlled amplifier 18 to increase, which will in turn increase the level of the output of the averaging circuit 20. Eventually, the system will stabilize with an amplification factor that is appropriate for the level of the input signal of the amplifier 18.
When a high voltage is applied by the averaging circuit 20 upon the voltage controlled amplifier 18, the voltage controlled amplifier 18 amplifies the phone line 12 signal least. A high voltage is applied by the averaging circuit 20 upon the voltage controlled amplifier 18 when a high signal is coming from the phone line 12. When the voltage controlled amplifier 18 is not amplifying due to the high signal voltage from the averaging circuit 20, the phone line 12 signal itself is the source of high signals and passes through the voltage controlled amplifier 18 relatively unaltered.
As low signals are amplified by the voltage controlled amplifier 18 while high signals are not amplified, the automatic gain controller 14 amplifies low, but not high signals. This tends to "pick up the bottom" of incoming signals from the phone line 12. When these signals are used as input to the ADC 10, low signals that would have experienced difficulty being represented digitally become more susceptible to digitization. The low signals are brought up to a higher level, medium signals are brought up to a somewhat higher level, and high signals are left at their original high level. The dynamic range of the incoming phone line 12 signal is compressed, making it more susceptible to digitization without quantization noise being as present upon playback.
The scaling circuit 16 present between the AGC 14 and the ADC 10 serves to mediate the signals from AGC 14 output for the ADC 10 input. The AGC 14 output voltage signal may be significantly less than or more than the optimum input signal level preferred by the ADC 10. In order to match the AGC 14 output signal with the ADC 10 input, a scaling circuit 16 is used that amplifies or diminishes the AGC 14 output by a constant factor for optimum ADC 10 input.
While the present invention has been described with regards to particular embodiments, it is recognized that additional variations of the present invention may be devised without departing from the inventive concept.