论文部分内容阅读
提出了一种结合 MBE(多带激励 )模型和 LPC(线性预测编码 )模型的 1 .8kbps声码器 .在这种声码器中 ,采用 LPC特征参数来代表语音帧的频谱 ,利用 LPC残差进行基音提取和多带清浊音判决 ,采用 MBE模型合成语音 ,并在高频浊音带的语音合成中混以清音 .在定点 Mo-torola DSP560 0 2 EVM上 ,可以在 1 .8kbps的码率下对语音进行实时的编解码处理 ,具有存储量和计算量较小的特点 .其合成语音质量超过了 LPC- 1 0 e.
A 1.8kbps vocoder combining MBE (Multi-band Excitation) model and LPC (Linear Predictive Coding) model is proposed.In this vocoder, LPC characteristic parameters are used to represent the frequency spectrum of speech frames, and LPC Poor for pitch extraction and multi-band voiced / unvoiced decision, MBE model for speech synthesis and mixed voicing in speech synthesis of high-frequency voiced speech bands. At fixed-point Mo-torola DSP560 0 2 EVMs, code rates of up to 1.8kbps Under the real-time encoding and decoding of voice processing, with a small amount of storage and computing characteristics of its synthetic voice quality exceeds LPC- 1 0 e.