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.

On 450-600 b/s natural sounding speech coding

Abstract: Algorithms for encoding speech with good intelligence and naturalness at very low rates are studied. Naturalness is retained by encoding accurately the speech excitation information from an LPC (linear predictive coding) model. A glottal ARX (autoregressive with exogenous input) technique is used to model the speech signal for high quality. A large reduction in coding rate is achieved through short-term temporal compression of the speech and vector quantization. Application of traditional vector quantization to the temporal decomposition output is discussed, with consideration of distortion measures and codebook generation. Based on properties of short-term temporal decomposition, finite-state vector quantization is introduced to further decrease the coding rate. A problem associated with this technique, estimation of a state transition matrix with incomplete data, is treated. The general result is that practical coders operating in a range of 450-600 b/s with a delay of about 200 ms and natural-sounding output speech can be designed. >

Interframe differential vector coding of line spectrum frequencies

Abstract: In many vocoders LSFs (line spectrum frequencies) are used to encode the linear predictive coding (LPC) parameters. An interframe differential coding scheme is presented for LSFs. The LSFs of the current speech frame are predicted by using both the LSFs of the previous frame and some of the LSFs of the current frame. Then the difference vector resulting from prediction is vector quantized. The proposed scheme is computationally efficient and easy to implement, and can be used in low-bit-rate vocoders. >

Arranging CELP information of one frame in a second packet

Abstract: Packets of real-time information are sent with a source rate greater than zero kilobits per second, and a time or path or combined time/path diversity rate initially being zero kilobits per second. This results in a quality of service QoS, optionally measured at the sender or the receiver. When the QoS is on an unacceptable side of a threshold of acceptability, the sender sends diversity packets at an increased rate. Increasing the diversity rate while either reducing or maintaining the overall transmission rate is new. CELP-based multiple-description data partitioning sends the base or important information plus a subset of fixed excitation in one packet and sends the base or important information plus the complementary subset of fixed excitation in another packet. Reconstruction produces acceptable quality when only one of the two packets is received and better quality when both packets are received. Reconstruction provides for single and multiple lost packets.