基于正弦模型的语音频域参数编码

本文提出了一种基于正弦模型的语音编码技术，通过对语音频率，幅值和相位参数的分析处理，合成高质量的语音。在编码处理过程中，我们应用了语音叠加技术和频迹跟踪技术，以提高合成语音清晰度，实验结果表明，谝编码方式具有很好的顽健性，适合于不同来语的信号，使如带背景音乐的语音。

A SPEECH CODING TECHNIQUE WITH FREQUENCY PARAMETER BASED ON SINUSOIDAL MODEL

In this paper, a speech coding technique based on sinusoidal model is developed. A high - quality speech waveform is obtained in analysis / synthesis processing with the frequencies, amplitudes, and phases of the speech. In addition, overlap - adding ( OLA) technique and frequency tracking technique are presented in order to improve the articulation of synthetic waveform. The experiment results indicate that this technique is provided with a good robustness and is fit for different sources of speech signal,such as speech in musical backgrounds.