宽带嵌入式语音编解码的帧擦除隐藏方法

Frame erasure concealment method used for wideband embedded speech codec

提出了一种宽带嵌入式编解码器的帧擦除隐藏方法。该方法在解码端将丢失帧分为静音、浊音、清音、清音向浊音的过渡、浊音向清音的过渡的语音类型,并根据语音类型对激励信号的能量采取对应的控制和调整。为了和宽带嵌入式编码器的结构相匹配,丢失帧的自适应码书根据前一帧的码流来恢复。为了增强编解码器的顽健性,在编码端采取了控制自适应码书贡献的技术。所建议的帧擦除隐藏技术不需要额外的比特和延迟,方法简单,恢复效果好,在提交给ITU-T的嵌入式变速率候选编码方案中得到使用。

An efficient frame erasure concealment （FEC） method for wideband embedded speech codec was proposed. The erased speech frame was classified as voiced, unvoiced, silence, unvoiced transit to voiced and voiced transit to unvoiced at decoder. The energy of excitations was carefully controlled based on the classification of the speech. To match with the configuration of the embedded speech codec, the adaptive codebook for erased frame was recovered with the last frame＇s bit-stream. For increasing the robustness of the codec, the contribution of adaptive codebook was propedy constrained at encoder. The proposed FEC method is very simple and has a good performance without extra delay and bits requirements in codec. This method has been applied to an embedded variable bits rate codec submitted to ITU-T as a candidate.