论文部分内容阅读
本文以多带激励声码器为模型，采用了多种技术去降低编码速率和改善音质.我们利用动态规划算法对基音周期进行平滑，去除了声码器中常有的音调噪声.MBE算法对谱包络的量化要花费大量的比特，这里利用LPC全极点模型谱逼近MBE谱包络，并采用共振峰增强技术来补偿模型误差.对谱幅度参数的量化，采用了分裂矢量量化（SPVQ）和多级矢量量化（MSVQ）的方法.使之在2.4kbps，1.2kbps及800bps等速率上，用单片ADSP2181实时实现了较高质量的语音压缩编解码.
In this paper, we use a multi-band excitation vocoder as a model, and adopt a variety of techniques to reduce the coding rate and improve the sound quality.We use the dynamic programming algorithm to smooth the pitch period and remove the common pitch noise in the vocoder.MBE algorithm Envelope quantification takes a large number of bits, where the LPC all-pole model spectrum is used to approximate the MBE spectral envelope and the formant enhancement technique is used to compensate for the model error. Splitting vector quantization (SPVQ) and Multi-level vector quantization (MSVQ) method to make it at 2.4kbps, 1.2kbps and 800bps equal rate, with a single ADSP2181 real-time high-quality voice compression codec.