Automatic gain control method and device for audio data

Abstract

The invention discloses an automatic gain control method and a device for audio data. The method comprises steps: framing processing is carried out on audio data, and feature parameters of each frame of data are extracted; according to the feature parameters of the current frame of data and a voice category Gaussian mixture model, the voice category condition probability of the current frame of data is obtained; according to the feature parameters of the current frame of data and a noise category Gaussian mixture model, the noise category condition probability of the current frame of data is obtained, and the possibility of the current frame of data belonging to a voice frame and the possibility of belonging to a noise frame are calculated; when the possibility of belonging to the voice frame is larger than that belonging to the noise frame, the current frame of audio data is judged to be the voice frame, and the gain is controlled according to a voice frame gain control rule; or otherwise, the current frame of data is judged to be the noise frame, and the gain is controlled according to a noise frame gain control rule. The technical scheme of the invention greatly improves the voice recognition level and the noise recognition level, automatic gain control is thus carried out, and the experience of a user can be improved effectively.

Description

technical field [0001] The invention relates to audio signal processing technology, in particular to an automatic gain control method and device for audio data. Background technique [0002] In the process of voice signal processing, the volume intensity of different audio signals is often different, accompanied by noise, but as a user, it is expected that the volume intensity of the call with everyone is the same without the volume key control Realize and improve user experience. The existing automatic gain control method analyzes the voice part and the noise part in the audio signal, and performs gain control on the two parts respectively. [0003] The existing automatic gain control methods all use time-domain analysis to distinguish speech and noise. This method of distinction has relatively large limitations and cannot effectively distinguish the characteristics of speech and noise. Often, speech will be recognized as noise, or noise will be recognized as noise. Recog...

AI Technical Summary

Problems solved by technology

[0003] The existing automatic gain control methods all use time-domain analysis to distinguish speech and noise. This method of distinction has relatively large limitations and cannot effectively distinguish the characteristics of speech and noise. Often, speech will be recognized as noise, or noise will be recognized as noise. Recognized as speech, resulting in incorrect gain control of the audio signal

For example, in a cochlear implant / hearing aid device, if the noise is mistakenly amplified, the user experience is very poor, and it may even cause serious discomfort to the user

Images