A wind noise detection and suppression method and apparatus suitable for speech and audio
A detection method and wind noise technology, applied in speech analysis, instruments, etc., can solve problems such as large amount of calculation, difficult to guarantee real-time performance, and no obvious directionality, so as to achieve wind noise, ensure voice quality and audio quality, avoid Effects of speech and audio distortion
Active Publication Date: 2019-01-15
北京声加科技有限公司
View PDF7 Cites 20 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] Wind noise is a strong non-stationary noise. At the same time, wind noise is not a point sound source and has no obvious directionality. Therefore, it is impossible to achieve effective wind noise by using traditional single-channel speech enhancement methods or multi-channel speech enhancement methods such as spatial filtering. detection and suppression
In recent years, some scholars have also proposed to use deep learning methods or methods based on non-negative matrix decomposition to realize wind noise detection and suppression [3]. Consumable communication application requirements, not universally applicable
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0019] The embodiment of the present invention proposes a low algorithm complexity real-time wind noise detection and suppression method, which is suitable for speech and audio signals, and can be applied to real-time speech and audio communication systems, and can also be applied to non-real-time speech and audio signal enhancement . In the embodiment of the present invention, by performing complex coherence function estimation on speech or audio, the sound source localization angles of multiple microphones in each speech frame are obtained in pairs, and then the sound source localization angle variance in the speech frame is obtained; Source localization angle variance determines the presence or absence of wind noise in speech frames.
[0020] In another embodiment of the present invention, the wind noise gain function in the speech or audio frame can be determined based on the sound source localization angle variance in the speech or audio frame; then, according to the wind...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention provides a wind noise detection and suppression method. In one embodiment, the method includes estimating a complex coherence function of speech or audio to obtain a sound source localization angle of a plurality of microphones in each speech frame, thereby obtaining a sound source localization angle variance in the speech frame; determining whether wind noise exist in a speech frameaccording to that angular variance of the sound source localization. Experiments show that the method can detect and suppress strong non-stationary wind noise in real time without obvious speech andaudio distortion.
Description
technical field [0001] The present application relates to the field of noise processing, in particular to a real-time wind noise detection and suppression method and device suitable for speech and audio. Background technique [0002] At present, the traditional single-channel speech enhancement method assumes the stationary characteristics of the noise, so the noise estimation method can be used to estimate the stationary noise power spectrum, and finally suppress the stationary noise [1]. The multi-channel speech enhancement method can take advantage of the spatial separation of target speech and interference noise, and perform spatial filtering through beamforming and other means to achieve stationary and non-stationary noise suppression [2]. The current method based on deep learning has a large amount of calculation, and the environmental adaptability and generalization ability need to be further improved. [0003] Wind noise is a strong non-stationary noise. At the same...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G10L21/0264G10L25/78
CPCG10L21/0264G10L25/78
Inventor 邱锋海匡敬辉
Owner 北京声加科技有限公司