Single-channel voice echo cancellation method and device based on deep neural network

The invention discloses a single-channel voice echo cancellation method and device based on a deep neural network. The method comprises the following steps: respectively calculating to obtain near-end and far-end frequency domain signals and extracting signal features; splicing the frequency domain...

Full description

Saved in:
Bibliographic Details
Main Authors ZHAO YUANJUN, YANG LIANG, GU CHENG
Format Patent
LanguageChinese
English
Published 03.01.2023
Subjects
ACOUSTICS
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
MUSICAL INSTRUMENTS
PHYSICS
SPEECH ANALYSIS OR SYNTHESIS
SPEECH OR AUDIO CODING OR DECODING
SPEECH OR VOICE PROCESSING
SPEECH RECOGNITION
Online AccessGet full text

Cover

More Information
Summary:The invention discloses a single-channel voice echo cancellation method and device based on a deep neural network. The method comprises the following steps: respectively calculating to obtain near-end and far-end frequency domain signals and extracting signal features; splicing the frequency domain signal features and then inputting the spliced frequency domain signal features to a coding frame, wherein the coding frame comprises three two-dimensional convolution layers, and a batch regularization layer and a PReLU layer which are located behind each two-dimensional convolution layer; inputting signal features output by the coding framework into a frequency GRU layer and a time GRU layer in sequence; inputting signal features output by the time GRU layer into a decoding frame; and outputting a voice time domain signal after carrying out optimization calculation on an output result of the decoding framework. According to the method, the correlation between the frequency points of the coding-decoding framework
Bibliography:Application Number: CN202211482692