Voice Signal Detection Using Wavelet Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting valid voice signals in noisy environments are inaccurate due to frequent noise interference, which can lead to incorrect identification of noise as the voice signal, especially in low signal-to-noise ratio conditions.
Innovation Solution
A method involving wavelet decomposition of audio signals to determine a valid voice signal by setting audio intensity thresholds based on maximum and minimum audio intensity values, distinguishing between noise and voice through specific threshold settings, and identifying valid voice segments by analyzing sample points with audio intensity values above these thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If energy-based voice detection is used in low signal-to-noise ratio conditions, then the detection method is simple, but the detection accuracy deteriorates due to noise being mistaken for voice signals
Solution Approach 1:
The audio signal is divided into multiple audio frames, and each frame is further decomposed into multiple wavelet decomposition signals. This segmentation allows the system to analyze local characteristics of the signal in different time-frequency regions, improving the ability to distinguish voice from noise compared to analyzing the entire signal as a whole.
Solution Approach 2:
The patent transforms the audio signal from the time domain to the time-frequency domain through wavelet decomposition. By analyzing the signal in multiple dimensions (different decomposition levels and time points), the system can better distinguish between voice and noise components that appear similar in the simple energy domain.
2Productivity
If simple energy comparison is used to detect voice signals, then the processing speed is fast, but the reliability deteriorates when noise frequently appears in adjacent time periods
Solution Approach 1:
The patent performs wavelet decomposition on each audio frame before energy comparison. This preliminary transformation into the time-frequency domain enables more reliable voice detection by revealing local signal characteristics that would be masked by noise in simple energy comparison, while still maintaining reasonable processing speed through efficient decomposition algorithms.
Data Source
AI summary
A method and apparatus for detecting a valid voice signal and a non-transitory computer readable storage medium are provided. A first audio signal including at least one audio frame signal is obtained. Multiple wavelet decomposition signals respectively corresponding to the at least one audio frame signal are obtained. A wavelet signal sequence is obtained by combining the multiple wavelet decomposition signals. A maximum value and a minimum value among audio intensity values of all sample points are obtained, and a first audio intensity threshold is determined according to the maximum value and the minimum value. Sample points each having an audio intensity value greater than the first audio intensity threshold in the wavelet signal sequence are obtained, and a signal of sample points in the first audio signal corresponding to the sample points each having an audio intensity value greater than the first audio intensity threshold is determined as the valid voice signal.