Hybrid Denoising Strategy for Multi-Microphone Audio Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise reduction techniques for speech processing in noisy environments, such as automotive vehicles and audio headsets, are inefficient when using a small number of microphones and require significant spatial separation, making them unsuitable for compact devices like automotive radios and headset earphones, especially at low frequencies where noise is most concentrated.
Innovation Solution
A hybrid denoising strategy using two sub-arrays of microphones, one for low frequencies and one for high frequencies, with distinct algorithms for each part of the spectrum: predicting noise in the low-frequency band and exploiting the predictable character of the useful signal in the high-frequency band, allowing for effective noise reduction with a small number of microphones in a compact configuration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional noise reduction techniques use a small number of microphones, then device complexity is reduced, but noise reduction effectiveness deteriorates
Solution Approach 1:
The patent divides the frequency spectrum into multiple bands (low-frequency band below 1kHz and high-frequency band above 1kHz) and applies different processing algorithms to each band. This segmentation allows effective noise reduction with fewer microphones by targeting the most problematic frequency range (low-frequency) with specialized prediction algorithms, while using standard processing for high frequencies where conventional methods work adequately.
2Reliability
If microphones are placed far apart to improve noise reduction, then noise discrimination capability is improved, but device compactness deteriorates
Solution Approach 1:
The patent changes the processing parameter approach by applying frequency-band-specific algorithms instead of relying solely on spatial separation. By using spectral prediction in the low-frequency band and exploiting signal predictability in the high-frequency band, the system achieves effective noise reduction with microphones placed close together, thus maintaining device compactness while preserving noise discrimination capability.
3Device complexity
If standard denoising algorithms are applied uniformly across all frequencies, then device complexity is reduced, but low-frequency noise reduction effectiveness deteriorates
Solution Approach 1:
The patent applies different processing qualities to different frequency regions: spectral prediction algorithms are applied specifically to the low-frequency band (below 1kHz) where noise is most concentrated and problematic, while standard processing is used for the high-frequency band (above 1kHz). This local quality approach optimizes noise reduction effectiveness in the critical low-frequency range without unnecessarily complicating the overall processing system.
Data Source
AI summary
This method comprises steps of: a) partitioning (10, 16) the spectrum of the noisy signal into a HF part and a LF part; b) operating denoising processes in a differentiated manner for each of the two parts of the spectrum with, for the HF part, a denoising by prediction of the useful signal from one sensor to the other between sensors of a first sub-array (R1), by means of a first adaptive algorithm estimator (14), and, for the LF part, a denoising by prediction of the noise from one sensor to the other between sensors of a second sub-array (R2), by means of a second adaptive algorithm estimator (18); c) reconstructing the spectrum by combining together (22) the signals delivered after denoising of the two parts of the spectrum, respectively; and d) selectively reducing the noise (24) by an Optimized Modified Log-Spectral Amplitude gain, OM-LSA, process.