Pick-up of Bone Conduction with Multiple Microphones and Quality Improvement of Bone Conducted Speech

针对骨导麦克风单次使用的情况，研究了骨导语音基频检测方法，提出了一种即使在高噪声环境下也能获得良好性能的检测方法。对骨导语音的幅值进行了时间域分析，发现将骨导语音的幅值转换为气导语音的幅值，可以提高语音质量。针对骨传导传声器多次使用的情况，采用自适应滤波技术构建了多输入单输出系统。研究了提供最佳语音质量改善的算法。

For the case of single use of a bone conducted microphone, fundamental frequency detection with bone conducted speech was investigated, and a detection method was derived, which provides a good performance even in highly noisy environments. Amplitude of bone conducted speech was investigated in the time domain, and it was found that a transform of amplitude of bone conducted speech into that of air conducted speech leads to an improvement of speech quality. For teh case of multiple use of bone conducted microphones, a multi-input single-output system was constructed involving adaptive filters. Algorithms to provide the best improvement of speech quality were investigated.