System and method for dual-microphone noise reduction applied to vehicle speech recognition

Abstract

The present invention relates to a system and method for dual-microphone noise reduction applied to vehicle speech recognition. The system comprises a main microphone, an auxiliary microphone, a speed mute module and a speed recognition subsequent processing module. The main microphone is configured to record passengers' sound and travelling noise; the auxiliary microphone is configured to record driving nose; the speed mute module is configured to process speech / noise signals recorded by the main microphone and the auxiliary microphone and optimize the speech / noise signals recorded by the auxiliary microphone through a noise model base to obtain noise signals, the processed main microphone speech / noise signals subtract the noise signals to obtain speed signals, the speed signals and a speed model base are subjected to matching optimization to obtain pure speed signals; and the speed recognition subsequent processing module is configured to identify sent speed signals and execute corresponding operation. The noise model base and the speed model base are arranged to match and optimize speed signals to obtain pure speed so as to improve the speech recognition rate in a vehicle and ensure the complete execution of a speed function.

Description

technical field [0001] The invention belongs to the technical field of automotive electronics applications, and in particular relates to a dual-microphone noise reduction system and a noise reduction method applied to vehicle-mounted speech recognition. Background technique [0002] With the development of technology in the automotive field, people's desire to use in-car electronic products is becoming stronger and stronger. [0003] Voice recognition is a new technology developed in recent years, and it is also a new function that has just penetrated into the automotive field. It controls some functions of the vehicle through voice interaction to enhance the driver's driving enjoyment. [0004] One of the key technologies of speech recognition is the clear processing of the sound quality of speech, that is, the noise reduction technology of speech. The existing noise reduction methods are as follows: [0005] Chinese patent CN201410076957.4 discloses a voice noise reducti...

AI Technical Summary

Problems solved by technology

[0005] Chinese patent CN201410076957.4 discloses a voice noise reduction method, using a single microphone scheme, calculating the estimated value of the noise power spectrum of the first 30 frames as a silent frame, calculating the average power spectrum when speaking as a voice frame, and subtracting the silent frame from the voice frame As the noise-reduced speech spectrum, the noise-reduced speech is obtained by processing. This method uses the noise before speaking as the silent frame, and does not use the background noise in real-time speaking as the silent frame. The processing of noise reduction is not accurate enough and cannot Used in high-precision functions such as speech recognition

[0006] Chinese patent CN201410042189.0 discloses a dual microphone speech noise reduction processing method and system. The dual microphone scheme is adopted, and the sum and average of the two signals are used as noise-containing speech signal estimation, and the difference average is used as noise signal estimation. The speech signal estimation is subtracted from the power spectrum of the noise signal estimation to obtain the pure speech signal power spectrum, and finally the pure speech signal is obtained by processing. The noise signal in this method is obtained by subtracting the two-way signal. Due to the noise of the vehicle Unlike human voice noise, so this method cannot be used in the car

Although this method is good at removing noise, it is not suitable for vehicle speech recognition

On the one hand, the main and auxiliary microphones of the vehicle are at a certain distance from the driver, and it is impossible to clearly distinguish noise from human voices. Secondly, the voice recognition function is also provided for the co-pilot and rear seat passengers, so this method cannot be used for vehicle voice recognition

Images