Audio Noise Suppression via Dynamic Channel Power Level Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio signal processing techniques face challenges in accurately estimating noise and sound power level differences between primary and reference channels, leading to suboptimal noise suppression and distortion in audio signals due to differences in microphone sensitivity and orientation.
Innovation Solution
The method involves estimating frequency-dependent Noise Power Level Difference (NPLD) and Speech Power Level Difference (SPLD) using maximum likelihood estimation and data-driven recursive noise power estimation, which corrects noise variance estimates and modifies filtering processes to account for these differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If noise suppression processing is applied using reference channel noise level estimates, then noise removal effectiveness is improved, but distortion of targeted sound occurs due to noise level differences between channels
Solution Approach 1:
The patent changes the noise level parameter from a fixed reference channel value to a dynamically adjusted value that accounts for NPLD. The system estimates the noise power level difference between primary and reference channels, then uses this difference to scale the reference noise variance estimate to match the primary channel's actual noise level, thereby improving noise suppression accuracy while preserving sound quality
Solution Approach 2:
The patent replaces the conventional direct copying of reference channel noise estimates with a statistical modeling approach. Instead of mechanically transferring noise variance values, the system uses probability density function modeling and maximum likelihood estimation to infer the primary channel's noise characteristics from the reference channel, accounting for systematic differences between channels
2Measurement precision
If reference microphone is positioned to capture background noise, then noise estimation capability is improved, but targeted sound leakage into reference channel causes estimation errors
Solution Approach 1:
The patent segments the reference channel signal analysis into noise-only periods and speech-containing periods. By identifying and separately processing these segments, the system can accurately estimate noise characteristics during noise-only periods while avoiding contamination from speech leakage, then apply these estimates to the overall noise suppression process
Solution Approach 2:
The patent performs preliminary noise estimation during noise-only periods before speech arrives. By capturing noise characteristics in advance during clean noise periods, the system establishes accurate baseline noise estimates that can be used during subsequent speech periods, avoiding the need to estimate noise in the presence of speech
Data Source
AI summary
A method for estimating a noise power level difference (NPLD) between a primary microphone and a reference microphone of an audio device includes obtaining primary and reference channels of an audio signal with primary and reference microphones of an audio device and estimating a noise magnitude of the reference channel of the audio signal to provide a noise variance estimate for one or more frequencies. A modelled probability density function (PDF) of a fast Fourier transform (FFT) coefficient of the primary channel of the audio signal is maximized to provide a NPLD between the noise variance estimate of the reference channel and a noise variance estimate of the primary channel. A modelled PDF of an FFT coefficient of the reference channel of the audio signal is maximized to provide a complex speech power level difference (SPLD) coefficient between the speech FFT coefficients of the primary and reference channel. A corrected noise magnitude of the reference channel is then calculated based on the noise variance estimate, the NPLD and the SPLD coefficient.