中文摘要：

类语音调制的目的是将输入数据比特流调制成具有类似语音的波形信号,以便于在话音通道上传输,实现数据抗声码器压缩传输。本文针对话音通道端到端的数据通信,提出了一种适用于多种声码器算法的类语音调制算法,该方法采用基于语料库的数据波形映射法,首先对TIMIT语料库进行预处理得到码本搜索空间,再利用增量搜索算法对其进行优化,最终生成最优码本。另外,对生成的最优码本进行了改进,使其能够更好地抗话音激活检测（Voice Activity Detector,VAD）。仿真结果表明,本文提出的方法在码本数目为16时,声码器AMR编码速率为12.2 kbps时符号错误率可达到0%,在声码器EFR、FR、G.729时符号错误率分别可达到0%、0%和0.0076%。

英文摘要：

The purpose of speech-like modulation research is to make the input data bit stream modulateto a set of signal that the features of it are like voice waveform in order to transform in the voice channeland implement the transmission of data for resistance to the vocoder. In view of the voice channel end-to-end data communication, this paper presents the algorithm of speech-like modulation which can applyto a variety of vocoders. The method adopts waveform data mapping based on corpus. First of all, itdeals with the corpus TIMIT to get the search space through the relevant pretreatment and usesincremental search algorithm to optimize it to generate optimal code. Besides, the optimal code is furtherimproved so that it is better to fight voice activity detector. Experimental results show that the proposedmethod can make the symbol error rate （SER） can get 0% in the rate of 12.2 kbps of AMR when thenumber of symbols is 16. The SER of EFR,FR,G.729 can get 0%,0% and 0.0076% respectively.