公开(公告)号：US20170061978A1

公开(公告)日：2017-03-02

申请号：US14536114

申请日：2014-11-07

申请人： SHANNON CAMPBELL

发明人： YUXUAN WANG ,  DELIANG WANG ,  KE HU

IPC分类号： G10L21/0208

CPC分类号： G10L21/0232 ,  G10L25/30

摘要： A method and system for separating noise from speech in real time is provided to improve intelligibility of speech for a variety of communications devices and hearing aids. From a speech signal, a plurality of frame-level features are extracted and form the input to the classifier. The classifier is a deep neural network comprising multiple hidden layers and an output layer with multiple output units. The classifier classifies the speech into a plurality of time-frequency units simultaneously. The classifier output constitutes an estimated ideal binary mask from which a fast gammatone filter bank is used to resynthesize the separated speech into an enhanced speech waveform.

摘要翻译： 提供一种用于实时分离语音噪声的方法和系统，以提高各种通信设备和助听器的语音清晰度。 从语音信号中，提取多个帧级特征并形成分类器的输入。 分类器是包含多个隐藏层的深层神经网络和具有多个输出单元的输出层。 分类器将语音分为多个时频单元同时进行分类。 分类器输出构成估计的理想二进制掩码，使用快速伽马网络滤波器组将分离的语音重新合成为增强语音波形。