论文部分内容阅读
针对高噪环境下语音识别的困难,提出一种基于独立分量分析的盲分离(ICA/BSS)与小波联合的语音降噪预处理方法,针对不同种类和不同输入信噪比的噪声设计了试验,结果表明基于ICA的语音识别预处理方法对低输入信噪比情况下多种噪声具有很强的鲁棒性和优越性,此结论对现实世界高噪环境下的信号分析和语音识别具有重要意义。
Aiming at the difficulty of speech recognition in noisy environments, a novel speech noise reduction preprocessing method based on ICA / BSS combined with wavelet is proposed. Experiments are designed for different kinds of noise with different input signal-to-noise ratios , The results show that ICA-based speech recognition preprocessing method is robust and superior to many kinds of noise under low input signal-to-noise ratio. This conclusion is important for signal analysis and speech recognition under the noisy environment of the real world significance.