论文部分内容阅读
以反映听觉特性的临界频带子波变换为工具,对混叠语音进行了分析,提出了一种基于临界频带子波变换的混叠语音源信号基音分离和检测方法,在此基础上利用数字多谐振荡器的选频特性,有效地获取构成混叠语音的每个语音信号的各谐波分量,以此实现从混叠语音中分离出各源信号。实验结果表明混叠语音分离方法是可行有效的。
Based on the analysis of the aliased speech with the critical frequency band wavelet transform which reflects the auditory characteristics, a method of pitch separation and detection of the aliased speech source signal based on the critical frequency band wavelet transform is proposed. On the basis of this, The frequency-selecting characteristic of the harmonic oscillator effectively obtains the harmonic components of each speech signal that constitutes the aliased speech, so as to separate each source signal from the aliased speech. Experimental results show that the method of separating speech is feasible and effective.