METHOD AND APPARATUS FOR AUTOMATICALLY CONTROLLING GAIN OF AUDIO DATA

WIPO Patent Application WO/2018/107874

A1

A method and an apparatus for automatically controlling the gain of audio data. The method comprises: dividing audio data into frames, and extracting the feature parameters of each frame of the data (S1); according to the feature parameters of the current frame of the data and a speech-type Gaussian mixed model, obtaining the speech-type conditional probability of the current frame of the data, and according to the feature parameters of the current frame of the data and a pre-configured noise-type Gaussian mixed model, obtaining the noise-type conditional probability of the current frame of the data (S2); according to the speech-type conditional probability of the current frame of the data and the noise-type conditional probability of the current frame of the data, calculating the probability of the current frame of the data being a speech frame and the probability of the current frame of the data being a noise frame (S3); when the probability of the current frame of the data being the speech frame is greater than the probability of the current frame of the data being the noise frame, determining the current frame of the audio data as the speech frame, and when the probability of the current frame of the data being the speech frame is less than the probability of the current frame of the data being the noise frame, determining the current frame of the data to be a noise frame (S4); when the current frame of the data is determined to be a speech frame, controlling the gain thereof on the basis of a pre-configured speech frame gain control rule, and when the current frame of the data is determined to be a noise frame, controlling the gain thereof on the basis of a pre-configured noise frame gain control rule (S5). The method can improve the level of identifying speech and noise, thereby automatically controlling the gain, and effectively improving user experience.

WO/2020/206975	METHOD FOR CALCULATING NUMBER OF SYLLABLES IN UNIT TIME AND RELATED APPARATUS
JP2002236499	MUSIC SIGNAL COMPRESSOR, MUSIC SIGNAL COMPANDER AND MUSIC SIGNAL PREPROCESSING CONTROLLER
JP2012531145	DSP-based device that aurally separates multi-sound inputs

LEI YANQIANG (CN)
CHENG XUEFENG (CN)

PCT/CN2017/104796

June 21, 2018

September 30, 2017

Click for automatic bibliography generation  

GUANGZHOU SHIYUAN ELECTRONICS CO LTD (CN)

G10L21/0272; G10L25/03; G10L25/51

CN106653047A	2017-05-10
CN105845150A	2016-08-10
CN101593522A	2009-12-02
CN105741849A	2016-07-06
CN103236260A	2013-08-07
CN105390142A	2016-03-09
CN101976566A	2011-02-16
US20100082339A1	2010-04-01
US5455888A	1995-10-03

SCIHEAD IP LAW FIRM (CN)

View/Download PDF      PDF Help

Previous Patent: VEHICLE NEGATIVE ION AIR PURIFIER

Next Patent: METHOD AND SYSTEM FOR MICROPHONE-CONNECTION LIVE BROADCAST OF A LIVE-BROADCAST PLATFORM