论文部分内容阅读
本文主要介绍一种新式声码器,这种声码器叫做剩余激励线性预测声码器《简称RELP声码器》。RELP声码器在概念上综合了线性预测编码(LPC)和话音激励声码两方面的优点。在RELP系统中,声道模拟是通过LPO技术来完成的,同时把LPO剩余信号作为激励信号。在低通滤波之后,采用自适应△调制对剩余信号进行编码,并在馈送到LPO合成器之前平坦其频谱。传输的速率范围通常是在6-9.6千比特/秒之间,这个范围内的合成语言质量是相当好的。当传输速率降低时,合成语言的质量下降得非常缓慢。由于不需要进行基音提取,因而这种声码器在任何工作场合下都很适用。
This article mainly introduces a new vocoder, called the residual excitation linear predictive vocoder “referred to as RELP vocoder.” The RELP vocoder conceptually combines the advantages of linear predictive coding (LPC) and voice-activated vocoding. In the RELP system, the channel simulation is done using the LPO technique, using the remaining LPO signal as the excitation signal. After low pass filtering, the remaining signal is encoded using an adaptive delta modulation and its spectrum flattened before being fed to the LPO synthesizer. The transmission rate is usually in the range of 6-9.6 kbit / s, and the quality of the synthesized language in this range is quite good. When the transfer rate is reduced, the quality of the synthesized language drops very slowly. This vocoder works well in any workplace because no pitch extraction is required.