Abstract: The invention uses an ATM Adaptation Layer of type 2 (AAL2) standard mechanism to define efficient support for Variable Rate Coding (VRC). The VRC in this context typically refers to codecs, which adapt their rate to information content variations in speech and audio. Such VRC results in lower average rate than the constant rate codecs or use of constant rate codecs coupled with silence suppression (SS), currently deployed in voice over ATM schemes using AAL2. Possible ATM transport, trunking and access applications encompass both Circuit Emulation Services (CES) and Local Loop Emulation Services (LLES). A typical VRC profile encompasses options for all sub-rates within one or more VRC standard or proprietary variable rate codec. The output of a rate determination algorithm (RDA), commonly part of variable rate codec, is fed into present AAL2 inter-working function (IWF).

Abstract: In methods and apparatus for encoding a gain parameter in a generalized linear predictive analysis-by-synthesis (GLPAS) coder, a subframe gain parameter is determined for each of a plurality of successive subframes of a frame, and a quantized frame gain parameter is determined for each frame using a delayed decision quantizer operating on the subframe gain parameters. The subframe gain parameters may be treated as components of a gain vector and the gain vector may be vector quantized to determine the quantized frame gain parameter. Encoder parameters are efficiently aligned with decoder parameters to ensure proper end-to-end operation. Alternatively, tree quantization or trellis quantization may be applied to the subframe gain parameters to determine the quantized frame gain parameter. The methods and apparatus are particularly applicable to low bit rate speech coding.

Abstract: In methods and apparatus for suppressing echo of a far end signal encoded using LPC-based compression in a near end signal encoded using LPC-based compression, parameters of each frame of the near end encoded signal are processed without synthesizing a speech signal from the near end encoded signal to determine whether sufficient echo to merit echo suppression is present in the frame. Upon determining that insufficient echo to merit echo suppression is present in the frame, the parameters of the frame are passed unmodified. Upon determining that sufficient echo to merit echo suppression is present in said frame, the parameters of the frame are modified without synthesizing a speech signal to suppress echo in the frame. The methods and apparatus are particularly suitable in codec bypass applications.

Abstract: In methods and apparatus for computing measures of echo of a far end signal in a near end signal, the evolution over time of frequency spectra of the near end and far end signals are compared to compute a measure of the echo. A far end spectrum of the far end signal and a near end spectrum of the near end signal are determined for each of a plurality of successive time intervals. A respective measure of correlation is determined for each of a plurality of spectrum pairs, each spectrum pair comprising a respective near end spectrum and a respective far end spectrum, the far end spectrum corresponding to a time interval which lags a time interval corresponding to the near end spectrum by a respective time lag. The measures of correlation are compared to determine a maximum measure of correlation which can be used as a measure of echo. The echo measure computation technique is particularly suitable for use in echo suppressors for digital cellular telephony systems, but has other applications.