使用动态分类器的用户语音活动检测
By using a dynamic classifier to perform adaptive clustering based on the frequency domain features of interaural phase and intensity differences, the problem of increased power consumption and resource requirements in existing technologies is solved, and efficient user voice activity detection is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- QUALCOMM INC
- Filing Date
- 2021-09-17
- Publication Date
- 2026-07-17
AI Technical Summary
While existing technologies improve accuracy in distinguishing between user voice and non-user voice, they also increase device power consumption and processing resource requirements, leading to reduced power efficiency.
A dynamic classifier is used for self-voice activity detection. Frequency domain features of interaural phase difference and intensity difference are extracted from the audio signals captured by the first and second microphones. Through adaptive clustering and decision boundary adjustment, user voice and other audio activities are distinguished.
Distinguishing user voice activity with high accuracy and relatively low power consumption under changing environmental conditions reduces device power consumption and processing resource requirements, thereby improving device power efficiency.
Smart Images
Figure CN116249952B_ABST