Abstract: In an encoder operable in response to a discrete pattern signal divisible into a succession of segments to produce an output code sequence, a pitch parameter and a spectral parameter are extracted in a parameter calculator from each segment and from a spectral interval. In an excitation pulse producing circuit, each spectral interval is divided into a plurality of subframes, namely, pitch periods with reference to the pitch parameter to divide each segment. A minor group of excitation pulses is calculated from the segment at every subframe to form a major group of the excitation pulses in the spectral interval. The excitation pulses of the major group are reduced in number with reference to adjacent ones of the minor groups in each spectral interval and are modified into a succession of modified excitation pulses. The modified excitation pulses are combined with the spectral parameter into the output code sequence.

Abstract: In an encoder operable in response to a discrete pattern signal divisible into a sequence of segments to produce an output code sequence, each segment is produced during a frame and specified by representative excitation signals extracted from each segment. The representative excitation signals may be representative pulses placed in a selected one of subframes formed by dividing the frame with reference to a spectral parameter and a pitch parameter extracted from each segment. Alternatively, the representative excitation signals may be either a combination of the representative pulses and a noise or a noise alone. The representative pulses and the spectral parameters may be subjected to interpolation. In a decoder for decoding the output code sequence into a reproduction of the discrete pattern signal, the representative pulses are interpolated to arrange excitation pulses in all of subframes of each frame and to produce an excitation vocal source signal.

Abstract: A voice encoding system is constituted by a short time voice signal series producing circuit inputted with a discrete voice signal series for dividing the same at each short time; a parameter extracting circuit for extracting a parameter representative of a spectrum envelope from the short time voice signal series and encoding the parameter; an impulse response series calculating circuit for calculating the impulse response series based on the parameter representative of the spectrum envelope; an autocorrelation function sequence calculating circuit utilizing the impulse response series; a cross-correlation function sequence calculating circuit utilizing the impulse response series and the short time voice signal series; a circuit for calculating and encoding an excitation signal series of the short time voice signal series by utilizing the autocorrelation function sequence; and a circuit for combining and outputting a code of the parameter representative of the spectrum envelope and a code representative of

Abstract: In a multi-channel system, a transmitting encoder includes an adaptive quantizer for each channel, an adaptive bit-allocator operating common to all channels using an update algorithm, and a multiplexer which rearrange signals responsive to bit-allocation information which in turn is not transmitted. At the receiving decoder, bit-allocation information is recovered by logic using the transmitter update algorithm. Four embodiments are disclosed.

Abstract: For use in combination with a loudspeaker and at least one microphone, for example, by attendants in an auditorium, an echo cancelling circuit comprises a self-adaptive echo canceller responsive to a lower frequency component, such as below 1.7 kHz, of a receive-in signal for self-adatively cancelling a corresponding component of a reverberation signal included in a send-in signal during each interval during which an audio signal reaches the circuit from a remote party. For a higher frequency reverberation signal component, an echo suppressor or a voice switch may reduce a weaker one of two signals which are either the higher frequency send-in and receive-in signal components or a combination of a reverberation component cancelled signal with the higher frequency send-in signal component and the whole receive-in signal. Alternatively, a less expensive echo canceller non-adaptively cancels a part of the reverberation signal in response to the receive-in signal.

Abstract: In an adaptive differential pulse code modulation (ADPCM) system for frequency band compression of speech or like signals, the coefficient of the synthesis filter in both the transmitter and receiver is varied in accordance with the normalized error e.sub.j /.DELTA. rather than the error itself, thus providing greater frequency band compression and preventing transmission errors from rendering the synthesis filter unstable.

Abstract: An adaptive differential pulse code modulation (ADPCM) system includes a predictor which predicts a sample value based on past prediction errors and coefficients which are adaptively corrected to lessen the difference, i.e. the prediction error, between the predicted values and the actual values. The predictor is duplicated in the receiver, has no feedback loop and thus instability due to transmission errors is eliminated. The system can also include a second predictor whose output is combined with that of the first predictor to obtain the predicted value. The second predictor output is based on past sums of the prediction error and the predicted value and coefficients which are adaptively corrected. The second predictor is in a feedback loop but instability is prevented by choosing the coefficients used.

Abstract: A device for developing a neutralization signal for an echo suppressor is required for the transmission of data over a telephone line having four-wire circuits. Each transmission of data is preceded by an identification signal at a given frequency. An input analog signal including the identification is sampled and quantized so that each sampled value is encoded into a binary codeword composed of a sign bit and a preset number of bits representing the magnitude of the analog signal. The device includes a detecting circuit to detect the identification signal, a level assessing circuit to assess the level of the codeword sequence, and a hangover circuit responsive to the detecting circuit and the level assessing circuit to develop the neutralization signal. The detecting circuit includes a bit pattern generator which produces successive bit patterns from sign bits of the codeword sequence.

Abstract: An adaptive hybrid circuit for connecting two-wire and four-wire telephone transmission lines to minimize singing, oscillation or echoes due to impedance mismatch. The circuit comprises a variable impedance circuit connected between the two-wire and four-wire circuits. In one embodiment one end of the variable impedance circuit is connected to the two-wire circuit and to one input of an adder to which the reception signal from the four-wire circuit is also applied and compared in reverse phase and half amplitude. The resulting transmission signal is used to adjust the variable impedance circuit so as to minimize the transmission signal. In another embodiment the variable impedance element is connected to an impedance through which the four-wire reception signal is applied and is further connected to one input of an adder which compares this input in reverse phase with the two-wire signal to produce a transmission signal.

Abstract: This data compression system further reduces previously-formed digital words representing samples of non-uniformly quantized and encoded analog speech, by normalizing and reencoding the original input digital word samples into smaller words using block coding wherein the input words are bit-reduced by normalizing a block of samples to the maximum amplitude sample value. Features of the invention include concatenating the most significant bit of the maximum value block code (first digital word) with the least significant bit of the segment bits part of the normalized input digital word (third digital word).

Abstract: An adaptive speech signal detector for use in a 4-wire telephone channel performs an adaptive threshold value setting operation depending on the channel noise level on a transmitter-side channel to detect a speech signal present at the transmitter. The adaptive operation of the speed signal detector is inhibited, however, if the signal level at the related receiver-side channel becomes higher than a preset value. This permits the use of the adaptive speech signal detector with DSI (digital speech interpolation) systems without malfunction due to the operation of an echo suppressor.

Abstract: Method and apparatus for cancelling a deviation in an input signal due to drift, line or background noise, changes in circuit component characteristics due to aging, and the like wherein the input signal, typically a digitized representation of an analog signal, is altered by a presumed offset magnitude; the polarity of the difference is monitored and a negative or positive count of clock pulses is accumulated dependent upon the aforesaid polarity. When a determined positive (or negative) count is reached the presumed offset is adjusted (up or down) by a predetermined increment and the count is begun anew.As an alternative technique and embodiment the initiation of the count may be restrained as long as the magnitude of the digitized input signal exceeds a predetermined threshold.

Abstract: A maximum value tracing circuit employed in echo suppressors and the like for selecting the peak values of digitized voice signals. The digitized value of a signal previously stored in a register is continuously compared with the present digitized value. The larger value of the compared signals is stored in the register after application of each clear signal to the register.First and second maximum value tracing circuits of the type described may be employed in combination with a comparator-selector circuit for comparing the outputs of the first and second maximum value extraction circuits to produce a comparison output signal representative of the larger of the two compared outputs derived from the first and second maximum value extraction circuits, which output signal is utiized for providing reliable, high speed echo suppression.

Abstract: An echo suppressor employed in systems interconnecting a two-wire circuit with a four-wire circuit for substantially eliminating "talker's echo" and for selectively attenuating the calling party's voice signal when the called party at the two-wire line begins speaking. The signal levels of signals on both paths of the four-wire circuit are measured to extract the maximum values of the signals for succeeding time intervals. The extracted value of the outgoing path signal is divided by the extracted value of the incoming path signal to generate an attenuation factor value. This value is multiplied by the value of the voice signal in the incoming path to generate an estimated leakage level signal. When this level is greater than the signal level in the output path (i.e. when there is no outgoing voice signal) the outgoing path is disconnected to prevent a talker's echo signal from being transmitted thereby.

Abstract: An echo controller comprises a parallel circuit of an echo suppressor and a self-adaptive echo canceller near an echo path of a long-distance telephone network. The echo controller further comprises a mode switch responsive to levels of a received signal, an unprocessed signal, and an output signal of the echo canceller and operable during absence of double talk for suspending operation of the echo canceller and putting instead the echo suppressor into operation only when characteristics of the echo path and operation of the echo canceller are objectionable.

Abstract: A speech detector which rapidly detects the presence of speech signals in telephone channel broadband noise is useful in TASI type systems. The incoming analog speech signal is sampled and binary-coded, then fed simultaneously to two different detectors: (1) A large-amplitude threshold detector (which detects high-energy vowel and plosive sounds): (2) a high-frequency threshold detector (which detects high-frequency fricative sounds above a 2 Khz threshold, assuming inherent noise in a 4 Khz telephone channel is mostly flat/broadband and 8 Khz sampling noise is filtered out).