Abstract: Properties of human audio perception are used to perform spectral and time masking to reduce perceived loudness of noise added to the speech signal. A signal is divided temporally into blocks which are then passed through notch filters to remove narrow frequency band components of the noise. Each block is then appended to part of the previous block in a manner which avoids block boundary discontinuities. An FFT is then performed on the resulting larger block, after which the spectral components of the signal are fed to a background noise estimator. Each frequency component of the signal is analyzed with respect to the background noise to determine, within various confidence levels, whether it is pure noise or a noise-and-signal combination. The frequency band's gain function is determined, based on the confidence levels. A spectral valley finder detects and fills in spectral valleys in the frequency component gain function, after which the function is used to modify the magnitude components of the FFT.

Abstract: A full duplex communications apparatus, having at each of a plurality of modem stations a transmitter and a receiver circuitry communicating with a two-wire system, receives signals having both a noise free signal component and a noise signal component. The noise signal component includes a far end echo signal which tends to corrupt and distort the received signal hence increasing the error rate. A method and apparatus for reducing the effect of the far end echo signal at the receiver, in particular a receiver using quadrature detection and equalization circuitry, provide a far end echo cancellation circuitry responsive to error signals generated by the receiver decision circuitry. The cancellation circuitry uses a synchronized reference signal from the transmitter of the station.

Abstract: Disclosed is a system and method for compressing data that ranks received data characters according to their frequency of occurrence in the preceding data stream. Each time a character is input to the system, the table of rankings of each character is updated through use of a bubbler system, which does a simple exchange between the input character and a character. After a significant portion of data has been read, the table will substantially reflect the relative ranks of the various characters. A set of probability pointers divides the various ranks into several ranges, and each time a new character is input and the rank of that character is altered, the probability pointers are also updated. In this fashion, the system can readily adapt to change in the type of data in the input stream. On the decompression side, similar tables are set up, with the ranks and the probability pointers constantly being updated after the receipt of each coded symbols.