摘要
要对染噪语音进行小波收缩以实现语音增强,首先需要考虑的问题就是小波分解中小波基的选择。工程人员一般按照经验选择小波基,并在相同应用目的中,使用相同的小波基,这种做法不利于小波收缩效果和稳健性。提出了一种基于消噪前后语音信号相异性的方法,在只有染噪信号的实际条件下,这种方法能自适应地选择出最佳小波基。试验结果证明,针对性地选择小波基有利于语音增强效果的提高。
In the implementation of wavelet shirnkage to speech enhancement, the first thing is to choose wavelet basis in wavelet decomposition. Usually, researchers choose wavelet basis according to their personal experiences and use the same basis for various application, which is adverse to speech enhancement. A new method to select best basis with only noisy speech possible is proposed, in which the dissimilarity between noisy and denoised signals is estimated. Simulation results show that the proposed method can find proper basis for speech enhancement, and thus bring better performance.
出处
《电声技术》
2008年第12期58-61,65,共5页
Audio Engineering
关键词
语音增强
小波收缩
小波基选择
信号相异性
speech enhancement
wavelet shrinkage
wavelet basis selection
signal dissimilarity
