提出了一种新的以基音周期为基础的分析方法。对语音信号，首先估计基音轨迹，即基音随时间的变化规律。然后根据这一轨迹对语音信号进行“基音调整”，将原始的、具有时变基音周期的信号转化为一个具有恒定基音周期的信号。这样就可以采用时不变的变换，例如ＭＬＴ（ｍｏｄｕｌａｔｅｄｌａｐｐｅｄｔｒａｎｓｆｏｒｍ）等。由于这些变换是基音同步的，其变换结果代表了语音的子带信号，具有明确的物理意义，便于后续的量化编码。这一分析方法应用于低数据率语音编码时，可以提高压缩比率或在同等数据率上提高合成语音的质量 A new method based on pitch period is proposed. For speech signals, first estimate the pitch trajectory, that is, pitch variation over time. Based on this track, the speech signal is then “pitch-adjusted” to convert the original, time-varying pitch cycle into a signal with a constant pitch. In this way, we can use the same transformation, such as MLT (modulatedlappedtransform) and so on. Since these transforms are pitch-synchronous, the transform result represents the sub-band signal of speech, with a clear physical meaning for subsequent quantization coding. This method of analysis, when applied to low data rate speech coding, can increase the compression ratio or improve the quality of the synthesized speech at the same data rate