Optical fiber speech enhancement method and system based on deep learning, medium and equipment

The invention discloses an optical fiber speech enhancement method, system, medium and device based on deep learning, and the method comprises the following steps: carrying out the short-time Fourier transform of a pure optical fiber speech signal and a noisy optical fiber speech signal, and corresp...

Full description

Saved in:
Bibliographic Details
Main Authors KWAK CHAN, CHAI SHIYI, CHEN HAO, CAO WEI, QIN HAIXIN
Format Patent
LanguageChinese
English
Published 26.05.2023
Subjects
ACOUSTICS
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 an optical fiber speech enhancement method, system, medium and device based on deep learning, and the method comprises the following steps: carrying out the short-time Fourier transform of a pure optical fiber speech signal and a noisy optical fiber speech signal, and correspondingly obtaining a pure optical fiber speech complex domain time-frequency signal and a noisy optical fiber speech complex domain time-frequency signal; performing complex proportion masking calculation on the pure optical fiber voice complex domain time-frequency signal and the noisy optical fiber voice complex domain time-frequency signal to obtain a target optical fiber voice complex proportion masking theoretical value; according to the noisy optical fiber voice complex number domain time-frequency signal and the target optical fiber voice complex number proportion masking theoretical value, based on a complex number convolutional neural network, obtaining a target optical fiber voice complex number domain ti
Bibliography:Application Number: CN202310203495