Improved adaptive CELP coder at 5.2 kbit/s
Shijun Yang, Xixian Chen
Abstract
Shijun Yang, Xixian Chen
Abstract
The code-excited linear predictive (CELP) coder is considered to be a good candidate for speech coding at low bit rates. Although the speech quality of this coder is good, certain distortions can still be perceived. In the paper, the authors introduce some procedures to improve the performance of the CELP coder. They improve synthetic speech quality quite a lot both in objective (signal-to-noise ratio) and subjective (listening tests) measures. Experimental results revealed that the proposed CELP coder improved segmental SNR by nearly 3 dB over the conventional CELP coder at 5.2 kbit/s.>
A significance statement is not available in the OpenAlex record.
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.
The code-excited linear predictive (CELP) coder is considered to be a good candidate for speech coding at low bit rates. Although the speech quality of this coder is good, certain distortions can still be perceived. In the paper, the authors introduce some procedures to improve the performance of the CELP coder. They improve synthetic speech quality quite a lot both in objective (signal-to-noise ratio) and subjective (listening tests) measures. Experimental results revealed that the proposed CELP coder improved segmental SNR by nearly 3 dB over the conventional CELP coder at 5.2 kbit/s.>
Key concepts: Code-excited linear prediction, Speech recognition, Computer science, Speech coding, Linear predictive coding, Vector sum excited linear prediction, Coding (social sciences), Mathematics