Code-excited linear prediction

About: Code-excited linear prediction is a research topic. Over the lifetime, 2025 publications have been published within this topic receiving 28633 citations. The topic is also known as: CELP.

Differential quantization of spectral parameters for CELP based coders in packet networks

Abstract: Line Spectrum Frequencies (LSF) have been the current parameters to represent LPC coefficients in speech coding. Extensive research has been performed to exploit their interframe and intraframe correlations and quantize them more efficiently. Interframe coding of LSFs can cause error propagation when frame erasures occur. Since most LSF quantizers were designed with the primary concerns of bit-rate and complexity, less attention was paid to error propagation. In this paper, we consider the erasure performance of LSF differential scalar quantizer (DSQ) and compare it with the interframe coding method embedded in the ITU-T G.729 coder. Our results show that with only 13,7% % extra bit-rate, DSQ algorithm is much more robust to frame erasures and improvements in terms of average spectral distortion (Avg. SD) and Enhanced modified bark spectral distortion (EMBSD) tests under various packet loss conditions are obtained.

Fast LSP vector quantization algorithms comparison

Abstract: The line spectrum pairs (LSP) provide an efficient representation of the synthesis filter used in linear predictive coding of speech. In this paper, an attempt to find the best distance measure for vector quantization is carried out, in the first place, by making objective studies over the same training sequence. Lastly, fast VQ algorithms of the LSP parameters are compared in terms of complexity, using the Euclidean distance measure. The well-known ordering property of LSP parameters is exploited to improve the efficiency of minimum distortion encoder for VQ in terms of norm associated to its distance. As conventional full search is too complex for practical implementation, the originality of this work consists in using the norm to limit the size of the area which contains the nearest neighbor of an input vector to he quantized. This method results in a substantial reduction in search complexity with only a minor degradation in terms of average spectral distortion. >

A 4800 bps CELP vocoder with an improved excitation

Abstract: The Stochastic or Code Excited Linear Predictive Coder (CELP) is among the promising candidates for producing good quality speech at low bit rates. However, the speech quality produced suffers from perceived roughness. Many researchers have used pole-zero postfilters to mask the roughness at the output of the synthesis filter. Although the postfilters are effective in masking the noise at low bit rates, they produce spectral distortions. It is proposed that speech can be improved by introducing two modifications to the fixed stochastic codebook. In the first modification, the stochastic codebook is used only when the long term correlations are low. Otherwise, a pulse like codebook is selected. In the second modification, the selected codebook output is weighted using an adaptive spectral shaping procedure. These two modifications were incorporated in a 4800 bps CELP coder and have resulted in a perceptually improved vocoded speech.

Study and subjective evaluation on MPEG-4 narrowband CELP coding under mobile communication conditions

Abstract: This paper evaluates MPEG-4 narrowband (NB) CELP speech coding under various mobile communication conditions, such as clean, background noise and transmission errors In order to make the codec robust against the errors with minimum increase of redundant bits, a CRC error correction code is attached into the codec as well as an error concealment is included in the decoder Subjective evaluation results demonstrate that the speech quality for MPEG-4 speech coding at above 83 kb/s is higher than that for the ITU-T G726 ADPCM at 32 kb/s in the clean speech condition Further, the speech quality degradation is less than 01 in MOS under 10/sup -3/ bit error conditions, and still comparable to or higher than that for G726 at 32 kb/s without error

Harmonic vector excitation coding of speech

Abstract: A coding algorithm for speech called harmonic vector excitation coding (HVXC) has been developed that encodes speech at very low bit rates (2.0–4.0 kbit/s). It breaks speech signals down into two types of segments: voiced segments, for which a parametric representation of harmonic spectral magnitudes of LPC residual signals is used; and unvoiced segments, for which the CELP coding algorithm is used. This combination provides near toll-quality speech at 4.0 kbit/s, and communication-quality speech at 2.0 kbit/s, thus outperforming FS1016 4.8-kbit/s CELP. This paper discusses the encoder and decoder algorithms for HVXC, including fast harmonic synthesis, time scale modification, and pitch-change decoding. Due to its high coding efficiency and new functionality, HVXC has been adopted as the ISO/IEC International Standard for MPEG-4 audio.