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.

A novel scheme for SVAC audio encoder

Abstract: In the audio encoder of Surveillance Video and Audio Coding (SVAC), both audio signals and MEL-frequency cepstral coefficients (MFCCs) are coded and this leads to high computational complexity. This paper proposes a novel scheme for SVAC in which speech coding module based on Algebraic Code Excited Linear Prediction (ACELP) is removed completely and speech waveforms can be reconstructed from MFCCs in decoding. The novel scheme greatly simplifies the structure of SVAC and also has a high performance for decoded speech signals in quality evaluation.

Predictive Neural Speech Coding

Abstract: Neural audio/speech coding has shown its capability to deliver a high quality at much lower bitrates than traditional methods recently. However, existing neural audio/speech codecs employ either acoustic features or learned blind features with a convolutional neural network for encoding, by which there are still temporal redundancies inside encoded features. This paper introduces latent-domain predictive coding into the VQ-VAE framework to fully remove such redundancies and proposes the TF-Codec for low-latency neural speech coding in an end-to-end way. Specifically, the extracted features are encoded conditioned on a prediction from past quantized latent frames so that temporal correlations are further removed. What's more, we introduce a learnable compression on the time-frequency input to adaptively adjust the attention paid on main frequencies and details at different bitrates. A differentiable vector quantization scheme based on distance-to-soft mapping and Gumbel-Softmax is proposed to better model the latent distributions with rate constraint. Subjective results on multilingual speech datasets show that with a latency of 40ms, the proposed TF-Codec at 1kbps can achieve a much better quality than Opus 9kbps and TF-Codec at 3kbps outperforms both EVS 9.6kbps and Opus 12kbps. Numerous studies are conducted to show the effectiveness of these techniques.

Adaptive codebook searching method based on correlation function in code-excited linear prediction coding

Abstract: PURPOSE: An adaptive codebook searching method based on a correlation function in code-excited linear prediction coding is provided to design an adaptive codebook with a scalar gain while having multi-tap structure CONSTITUTION: A past excite signal is modeled through a second-order smoothing filter A factor that minimizes the sum of squares of error signals is calculated An excite signal is obtained using the calculated factor A pitch delay and a gain value are obtained from the excite signal A gain vector is estimated from the past excite signal and a three-dimensional gain vector is represented in scalar from the estimated value Multi-gain is modeled into single-gain using correlation of the excite signal

The 2003 Benjamin Franklin medal in electrical engineering presented to Bishnu S. Atal

Abstract: Dr. Atal was chosen for his fundamental contributions to speech coding in digital cellular telephony—specifically for pioneering work in linear predictive coding (LPC) and use of analysis-by-synthesis leading to multi-pulse excited linear predictive coding and code excited linear predictive coding (CELP). These are the basis of time-division and code-division cellular systems (TDMA & CDMA), the Nextel system, and the global system for mobile communications (GSM).