All deep learning minimum variance distortionless response beamformer for speech separation and enhancement
AssigneeTencent America LLC
InventorsYong Xu, Meng Yu, Shi-Xiong Zhang, Dong Yu
Life of this patent
Estimating the term…
Abstract
A method, computer program, and computer system is provided for automated speech recognition. Audio data corresponding to one or more speakers is received. Covariance matrices of target speech and noise associated with the received audio data are estimated based on a gated recurrent unit-based network. A predicted target waveform corresponding to a target speaker from among the one or more speakers is generated by a minimum variance distortionless response function based on the estimated covariance matrices.
Claims17
A method of speech recognition, executable by a processor, comprising:
- receiving audio data corresponding to one or more speakers;
- estimating covariance matrices of target speech and noise associated with the received audio data based on a gated recurrent unit-based network (GRU-Net);
- generating a predicted target waveform corresponding to a target speaker from among the one or more speakers by a minimum variance distortionless response (MVDR) function based on the estimated covariance matrices;
- and
- recursively accumulating and updating the covariance matrices by the GRU-Net for one or more frames.
A computer system for speech recognition, the computer system comprising:
- one or more computer-readable non-transitory storage media configured to store computer program code;
- and
- one or more computer processors configured to access said computer program code and operate as instructed by said computer program code, said computer program code including:
- receiving code configured to cause the one or more computer processors to receive audio data corresponding to one or more speakers;
- estimating code configured to cause the one or more computer processors to estimate covariance matrices of target speech and noise associated with the received audio data based on a gated recurrent unit-based network (GRU-Net);
- generating code configured to cause the one or more computer processors to generate a predicted target waveform corresponding to a target speaker from among the one or more speakers by a minimum variance distortionless response (MVDR) function based on the estimated covariance matrices;
- and
- accumulating code and updating code configured to cause the one or more computer processors to recursively accumulate and update the covariance matrices by the GRU-Net for one or more frames.
A non-transitory computer readable medium having stored thereon a computer program for speech recognition, the computer program configured to cause one or more computer processors to:
- receive audio data corresponding to one or more speakers;
- estimate covariance matrices of target speech and noise associated with the received audio data based on a gated recurrent unit-based network (GRU-Net);
- generate a predicted target waveform corresponding to a target speaker from among the one or more speakers by a minimum variance distortionless response (MVDR) function based on the estimated covariance matrices;
- and
- recursively accumulate and update the covariance matrices by the GRU-Net for one or more frames.
Dependencies are inferred from the claim text.