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.

Speech encoding device and decoding device

Abstract: PROBLEM TO BE SOLVED: To reduce memory capacity by using a vector series outputted from an oscillator as fixed vectors or probabilistic vectors. SOLUTION: This device is equipped with a seed storage part 11 which stores seeds, an oscillator 12 which outputs different vector series corresponding to the values of the seeds stored in the seed storage part 11, and an LPC synthesizing filter part 13 which inputs the vector series as sound source vectors and performs LPC synthesis to obtain a synthesized sound. This is obtained by replacing the fixed vector readout part and fixed code book of a conventional CELP type speech encoding device or the fixed vector readout part and fixed code book of a CELP type speech decoding device with the oscillator 12 and seed storage part 11. Then the oscillator 12 outputs different vector series corresponding to the values of inputted seeds. Further, the LPC synthesizing filter part 13 performs the LPC synthesis of inputted sound source vectors 15 and outputs a synthesized sound 16.

Intraframe quantization of speech line spectrum pairs for code-excited linear prediction based coders in packet networks

Abstract: This paper proposes intraframe quantization schemes of line spectrum pair parameters to improve frame erasures for the ITU-T G.723.1 coder. The standard ITU-T G.723.1 coder uses an interframe quantization of the line spectrum pair parameters, which causes error propagation to the next frames. Simulation results show that our intraframe quantization schemes coding is much more robust to frame erasures than the embedded method in ITU-T G.723.1, and a typical improvement of 0.4 dB on average spectral distortion can be obtained with 40% packet loss. Enhanced modified bark spectral distortion tests, under various packet loss conditions, confirm that our proposed method is superior to the interframe algorithm embedded in the ITU-T G.723.1 coder of over 1.6. Copyright © 2012 John Wiley & Sons, Ltd.

Combined speech and audio coding using non-linear adaptations

Abstract: A combined speech and audio coder is proposed. The coder structure resembles a low-delay CELP coder, however, the excitation gain is adapted non-linearly in a sample-by-sample fashion by using a trained neural network, and the spectral parameters are derived from backward non-linear prediction based on a second-order Volterra filter. A perceptual weighting filter derived from psychoacoustic analysis in the spectral domain is used to shape the coding noise. The proposed non-linear adaptation schemes significantly improve the effectiveness of using an analysis-by-synthesis model for coding audio signal. Simulation results show that transparent coding of wideband (7 kHz) speech and audio at 24 kbps is achieved.

Natural sounding speech coder operating at 2.4 kb/s and below

Abstract: Accurate representation of periodic speech segments is essential for synthesizing high quality digital speech at very low bit rates For bit rates at 48 kb/s and below, conventional code excited linear prediction (CELP) does not provide the appropriate degree of periodicity It has been suggested that good quality low bit rate speech can be obtained from frequency domain techniques such as STC, NMR and MB-LPC vocoders A MB-LPC vocoder operating at 24 kb/s and below is presented which exploits the advantages of both time- and frequency-domain speech coding to produce natural sounding, good quality speech This also includes a new frequency domain post-filter which attenuates the noise in the formant nulls and enhances the synthesized speech >

On a reduction of pitch searching time by preprocessing in the celp vocoder

Abstract: Code Excited Linear Prediction(CELP) speech coders exhibit good performance at data rates below 4.8 kbps. This major drawback of CELP type coders is required much computation. In this paper, we propose a new pitch search method that preserves the quality of the CELP vocoder with reducing complexity. In the pitch searching, we detect the segments of high correlation by a simple preprocessing, and then carry out the pitch searching only for the segments obtained by the preprocessing. By using the proposed method, we can get approximately complexity reduction in the pitch search.