Transient Noise Detection via Wavelet Energy Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing transient noise detection methods in audio technology suffer from low accuracy due to the similarity in energy change characteristics between speech onset and transient noise, leading to incorrect identification of speech as noise.
Innovation Solution
A method involving wavelet decomposition of audio frame signals to determine energy distribution information and calculate probabilities of transient noise, using sub-wavelet signals to refine noise detection in a more precise time dimension, and subsequent noise suppression techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If transient noise detection analyzes signal energy sharp increase in a period of time, then transient noise can be detected, but speech onset is incorrectly identified as noise due to similar energy change characteristics
Solution Approach 1:
The patent segments the audio signal into multiple frames and further decomposes each frame into sub-frames using wavelet transform. This segmentation allows the system to analyze energy changes at different temporal resolutions, distinguishing between short-duration transient noise and longer speech onsets. By examining energy distribution across multiple segmented time intervals, the system can accurately identify transient noise while avoiding false detection of speech.
Solution Approach 2:
The patent introduces a frequency domain dimension by applying wavelet transform to the time-domain audio signal. This transforms the signal into time-frequency representation, allowing analysis of energy distribution not only in time but also in frequency. The system calculates energy distribution across different frequency bands and time periods, creating a multi-dimensional feature space that effectively distinguishes transient noise from speech onset characteristics.
2Productivity
If the system detects sharp energy changes to identify transient noise, then noise detection capability is achieved, but the beginning of audio signal (speech occurrence) is also detected as noise
Solution Approach 1:
The patent applies different detection criteria to different local regions of the audio signal. By using wavelet decomposition, the system analyzes local energy distribution characteristics within specific time-frequency cells. The system calculates local energy distribution and compares it with global characteristics, allowing adaptive detection that accounts for local signal properties. This enables accurate distinction between transient noise (with specific local energy patterns) and speech onset (with different local characteristics).
Solution Approach 2:
The patent changes the detection parameters from simple energy thresholding to a multi-parameter assessment including energy distribution, temporal characteristics, and frequency content. The system calculates energy distribution across multiple time periods and frequency bands, and uses these distributed parameters to make detection decisions. This parameter transformation allows the system to maintain high noise detection capability while improving precision by examining the pattern and distribution of energy changes rather than relying on single-point threshold crossing.
Data Source
AI summary
Disclosed is a method, an apparatus, and a device for transient noise detection. The method includes: obtaining an audio frame signal having a preset duration; performing wavelet decomposition on a first audio frame signal to obtain a first wavelet decomposition signal corresponding to the first audio frame signal; determining a first reference audio intensity value of a first sub-wavelet decomposition signal according to reference audio intensity values of all samples in the first sub-wavelet decomposition signal; determining energy distribution information of the first wavelet decomposition signal according to first reference audio intensity values of all sub-wavelet decomposition signals in the first wavelet decomposition signal; and determining a probability that the first audio frame signal is transient noise according to the energy distribution information of the first wavelet decomposition signal.