Speech communication system and method for handling lost frames

Abstract

A speech communication system and method with an improved method for handling lost information during transmission from an encoder to a decoder. More specifically, the improved speech communication system can more accurately recover lost information about a speech frame, such as line spectral frequency (LSF), pitch lag (or adaptive codebook excitation), fixed codebook excitation and/or or gain information. To handle missing LSFs, the improved speech communication system sets the minimum spacing between LSFs to an increased value, and then optionally decreases this value for subsequent frames in a controlled adaptive manner. To deal with a missing pitch lag, the improved system estimates the pitch lag for the missing frame by extrapolating from the pitch lags of previously received frames. When the improved decoder receives the pitch lag of a subsequently received frame, the system uses a curve fit between the pitch lag of the previously received frame and the pitch lag of the subsequently received frame in order to fine-tune its response to the An estimate of the pitch lag of the lost frame is used to adjust and correct the adaptive codebook buffer before it is used by subsequent frames. In dealing with loss gain parameters, the improved system estimates loss gain parameters depending on whether the speech is periodic or aperiodic, whether the loss gain parameters are adaptive codebook gain parameters or fixed codebook gain parameters, and other factors , such as the average adaptive codebook gain parameter of an adaptive number of subframes of previously received frames, the ratio of the adaptive codebook excitation energy to the total excitation energy, the spectral tilt of previously received frames and/or the previously received frame energy. If the speech communication system does not send fixed codebook excitation values ​​to the decoder, the improved encoder/decoder generates the same random excitation values ​​for a given frame using a seed value whose value is determined by information in that frame. After estimating loss parameters in a lost frame and synthesizing speech, the improved system matches the energy of the synthesized speech to the energy of a previously received frame.

Description

[0001] References [0002] The following U.S. patent applications are hereby incorporated by reference in their entirety and made a part of this application: [0003] U.S. Patent Application Serial No. 09 / 156,650, "Speech Encoder Using Gain Normalization That Combines Open AndClosed Loop Gain," filed September 18, 1998, Conexant Docket No. 98RSS399; [0004] U.S. Provisional Application Serial No. 60 / 155,321, "4 kbits / s Speech Coding," Conexant Docket No. 99RSS485, filed September 22, 1999; and [0005] US Patent Application Serial No. 09 / 574,396, "A New Speech Gain Quantization Strategy," filed May 19, 2000, Conexant Docket No. 99RSS312. Background technique [0006] This invention relates generally to the encoding and decryption of speech in voice communication systems, and more particularly to methods and apparatus for handling erroneous or lost frames. [0007] To model basic speech, the speech signal is sampled in time and stored frame by frame as discrete waveforms to ...

AI Technical Summary

Problems solved by technology

However, if a noisy frame is lost and not received by the decoder, the encoder and decode

Images