摘要
提出一种基于接收端的联合语音解码算法,根据信道或信道解码提供的参数软输出值,以及语音编码参数的冗余量计算接收参数的后验概率;利用后验概率基于MAP或MS准则进行参数估计,最佳估计值进行语音解码.仿真中采用G.711和GSMEFR两种语音编码标准,结果表明联合解码算法显著提高了解码语音信噪比.
A joint speech-decoding algorithm is proposed in this paper. Based on the soft-value of channel decoder and the redundancy of source parameter, a posterior probability is calculated and used to estimate speech parameter according to MAP/MS criteria. Next the optimal parameter from the estimation model is decoded. Simulation carried our in accordance with G.711 and GSM EFR speech coding standard shows that the joint decoding algorithm obviously improves the SNR of decoded speech.
出处
《应用科学学报》
CAS
CSCD
北大核心
2005年第3期240-244,共5页
Journal of Applied Sciences
