摘要
大部分传统的语音增强算法不能提高语音的可懂度,同时还引入了失真。要想提高语音的质量和可懂度,需要设计相应的准则保留衰减失真的有用信道,消除放大失真的不利信道。利用基于统计模型的方法结合改进的IMCRA噪声估计算法对语音进行降噪,再使用SNR_(ESI)信道选择标准保留有用信道、消除不利信道,对通过有用信道的语音进行合成达到语音增强的目的。实验结果表明,在不同信噪比输入下的非平稳噪声环境中,该算法能够有效提高语音的质量和可懂度。
Most of the traditional speech enhancement algorithms can't improve the intelligibility of speech,and at the same time,they introduce distortion.In order to improve the quality and intelligibility of speech,it is necessary to design corresponding criteria to retain the useful channel of attenuation distortion and eliminate the unfavorable channel of amplification distortion.The method based on statistical model was combined with the improved IMCRA noise estimation algorithm to reduce the noise of speech.The SNR_(ESI)channel selection standard was used to retain the useful channel and eliminate the adverse channel.The speech of the useful channel was synthesized to achieve the purpose of speech enhancement.The experimental results show that the algorithm can effectively improve the quality and intelligibility of speech in non-stationary noise environments with different signal-to-noise ratio inputs.
作者
卢勇
舒涛
Lu Yong;Shu Tao(Air Traffic Management Center,Civil Aviation Flight University of China,Guanghan 618307,Sichuan,China)
出处
《计算机应用与软件》
北大核心
2022年第11期148-153,共6页
Computer Applications and Software
基金
青年基金项目(Q2019-072)。
关键词
语音增强
语音可懂度
信道选择
信噪比
Speech enhancement
Speech intelligibility
Channel selection
Signal-to-noise ratio
