论文部分内容阅读
首先介绍了帧同步混合小波包的分析方法。该方法结合了小波包时频窗口可变的特点和STFT的分帧处理形式。它既能够保证语音信号处理中帧长的要求,即可实时处理,义能获得对信号频域上的最佳分解,是一种类似FFT的小波包的快速算法。在此基础上,应用该方法模拟了听觉模型,并运用于语音增强。实验表明,即使在-5 dB低信噪比的条件下,也能获得良好的除噪效果和听觉效果。该方法还可运用于语音的编码、合成和识别等领域。
First introduced the frame synchronization hybrid wavelet packet analysis. The method combines the characteristics of wavelet packet time-frequency window variable and STFT framing processing. It not only guarantees the frame length requirement in speech signal processing, but also realizes the real-time processing. It can obtain the best decomposition in the signal frequency domain, which is a fast algorithm similar to FFT wavelet packet. On this basis, this method is used to simulate the auditory model, and applied to speech enhancement. Experiments show that, even at -5 dB low signal-to-noise ratio conditions, but also get a good noise removal effect and auditory effect. The method can also be applied to speech coding, synthesis and identification and other fields.