摘要
MELP(多带混合激励线性预测)是一种高效实用的语音编解码器.在2.4 Kb/s速率上,语音质量基本达到4.8 Kb/s FS-1016 CELP声码器质量,在抗噪性能方面甚至比CELP还高,接近16 Kb/s SVSD编码器.对2.4 Kb/sMELP编码器帧处理方面做一些改进,将过渡帧划分为2个子帧,分别进行编码和合成,使MELP编码器得到更高的语音质量.
JP2MELP（ mixed-excitation linear predictive） is a high-efficiency speech codec, which is developed on the basis of the LPC codec. At the speed of 2.4 Kb/s, the speech quality of the MELP is approaching that of 4. 8Kb/s FS - 1016 CELP on which the speed is 4. 8 Kb/s, while the noise-resistant ability is higher than that of CELP,approaching 16 Kb/s SVSD codec. To improve the performance of the 2. 4Kb/s MELP codec,the transition frame should be cut into two smaller frames, so as to achieve the higher speech quality.
出处
《应用科技》
CAS
2006年第6期11-13,共3页
Applied Science and Technology
