Multipulse excitation codebook design and fast search methods for CELP speech coding
F.F. Tzeng
Abstract
F.F. Tzeng
Abstract
Presents an efficient method for multipurpose excitation coding using direct vector quantization. This method is then extended to the design of the excitation codebook for a CELP (code excited linear predictive) coder. For the best excitation-vector search at low data rates, an approach combining direct vector quantization and an analysis-by-synthesis procedure is used. Computer simulation results for a 4.8 kb/s speech coder based on the new excitation codebook are presented.>
OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Presents an efficient method for multipurpose excitation coding using direct vector quantization. This method is then extended to the design of the excitation codebook for a CELP (code excited linear predictive) coder. For the best excitation-vector search at low data rates, an approach combining direct vector quantization and an analysis-by-synthesis procedure is used. Computer simulation results for a 4.8 kb/s speech coder based on the new excitation codebook are presented.>
Key concepts: Codebook, Code-excited linear prediction, Linde–Buzo–Gray algorithm, Vector quantization, Vector sum excited linear prediction, Linear predictive coding, Speech coding, Computer science