Adaptive IIR Filter for Audio SNR Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio processing devices, such as hearing aids, face challenges in accurately estimating the signal-to-noise ratio (SNR) in noisy environments, particularly in conditions with varying noise levels and speech onsets, leading to suboptimal noise reduction and increased musical noise.
Innovation Solution
The implementation of a recursive algorithm using a 1st order IIR low-pass filter with an adaptive cut-off frequency, based on both a posteriori and a priori SNR estimates, for non-linear smoothing to improve noise reduction, especially in low SNR conditions, and a Directed Bias and Smoothing Algorithm (DBSA) to adjust smoothing and bias parameters dynamically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If non-linear smoothing with adaptive low-pass filtering is applied to a posteriori SNR estimate, then noise reduction performance is improved, but computational complexity increases
Solution Approach 1:
The patent implements dynamic adaptation of the low-pass filter cut-off frequency based on the a posteriori SNR estimate. The cut-off frequency is adjusted in real-time according to noise conditions, allowing the system to optimize between smoothing strength and responsiveness. This dynamic parameter adjustment resolves the contradiction by making the computational load adaptive rather than fixed, intensifying processing only when noise conditions warrant it.
Solution Approach 2:
The patent changes the parameter being filtered from the raw a posteriori SNR estimate to a non-linearly smoothed version derived through adaptive low-pass filtering. This parameter transformation improves reliability by producing a more stable a priori SNR estimate that better reflects true noise conditions, while the adaptive nature of the filtering keeps computational complexity manageable.
2Object-generated harmful factors
If aggressive smoothing is applied to SNR estimates, then musical noise is reduced, but speech clarity deteriorates
Solution Approach 1:
The patent dynamically adjusts the smoothing aggressiveness based on the a posteriori SNR estimate. In low SNR conditions where musical noise is more problematic, stronger smoothing is applied. In high SNR conditions where speech clarity is already good, less smoothing is applied. This dynamic adjustment resolves the contradiction by making smoothing intensity context-dependent rather than uniform.
Solution Approach 2:
The patent applies different smoothing characteristics to different regions of the SNR estimate based on local noise conditions. The adaptive low-pass filter effectively creates locally optimized smoothing behavior, applying stronger smoothing where needed and weaker smoothing where speech clarity is already maintained, thus resolving the contradiction between noise reduction and speech preservation.
3Ease of manufacture
If fixed cut-off frequency low-pass filtering is used, then implementation is simple, but adaptation to changing noise environments is slow
Solution Approach 1:
The patent transforms a static filter design into a dynamic one by making the cut-off frequency a function of the a posteriori SNR estimate. This allows the filter to automatically adapt to changing noise environments without requiring complex manual configuration or multiple fixed filters. The implementation remains relatively simple while gaining significant adaptability through this single dynamic parameter.
Solution Approach 2:
The patent changes the cut-off frequency parameter of the low-pass filter based on the estimated noise conditions. This parameter change enables the system to adapt to varying SNR environments, transitioning from a fixed implementation to an adaptive one while maintaining implementation simplicity through the use of a single adjustable parameter rather than multiple complex filters.
Data Source
Figure 1A~2
Figure 3~4
Figure 5
AI summary
The application relates to an audio processing device comprising a) at least one input unit for providing time-frequency representation Y(k,n) of an electric input signal representing a time variant sound signal consisting of target speech signal components S(k,n) from a target sound source TS and noise signal components N(k,n) from other sources than the target sound source, where k and n are frequency band and time frame indices, respectively, b) a noise detection and/or reduction system configured to b1) determine an a posteriori signal to noise ratio estimate γ(k,n) of said electric input signal, and to b2) determine an a priori target signal to noise signal ratio estimate ζ(k,n) of said electric input signal from said a posteriori signal to noise ratio estimate γ(k,n) based on a recursive decision directed algorithm. The a priori target signal to noise signal ratio estimate ζ(k,n) for the nth timeframe is determined from the a priori target signal to noise signal ratio estimate ζ(k,n-1) for the (n-1)th timeframe and the a posteriori signal to noise ratio estimate γ(k,n) for the nth timeframe. The application further relates to a method of of estimating an a priori signal to noise ratio. Thereby improved noise reduction may be provided. The invention may e.g. be used for the hearing aids, headsets, ear phones, active ear protection systems, handsfree telephone systems, mobile telephones, etc.