Adaptive IIR Noise Suppression for Hearing Aid Input Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise suppression methods in hearing devices amplify noise components due to the natural speech rhythm, leading to undesirable broadband sound artifacts, and struggle to maintain the signal-to-noise ratio effectively with limited processing resources.
Innovation Solution
A method that uses a filter bank to divide the input signal into frequency bands, estimates the frequency and spectral width of useful signal components, and applies adaptive filtering with IIR filters to suppress noise while preserving the signal level, utilizing a complex pole transfer function and variance-based parameters to minimize co-modulation of noise with speech components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If frequency band-dependent amplification is applied to suppress noise, then the signal-to-noise ratio is improved, but noise components are comodulated with speech components leading to audible broadband sound artifacts
Solution Approach 1:
The patent applies different amplification factors to different frequency bands based on local signal characteristics. By analyzing the spectral content of each frequency band and applying targeted amplification only where speech components are present, the method achieves noise suppression without comodulating broadband noise artifacts, as each band is processed independently according to its local quality requirements
Solution Approach 2:
The patent segments the frequency spectrum into multiple frequency bands using a filter bank. This segmentation allows independent processing of each band, enabling the system to identify and amplify speech components in specific bands while suppressing noise in other bands, thereby avoiding the comodulation of broadband noise that occurs when applying uniform amplification across the entire spectrum
2Reliability
If the amplification factor is increased to suppress noise in frequency bands without useful signal, then noise suppression is improved, but the useful signal level is also reduced when it occurs in different frequency bands
Solution Approach 1:
The patent dynamically adjusts the amplification factor for each frequency band based on real-time spectral analysis. The system continuously monitors the presence and level of speech components in each band and adapts the amplification accordingly, ensuring that noise is suppressed in bands without speech while maintaining appropriate signal levels in bands containing useful speech components
Solution Approach 2:
The patent employs feedback mechanisms where the spectral content of each frequency band is analyzed and used to control the amplification factor for that band. This feedback loop ensures that the amplification decisions are based on actual signal characteristics, preventing both excessive noise suppression and unnecessary reduction of useful signal components
Data Source
Figure 1~2
Figure 3~4
AI summary
A method (1) for frequency-dependent noise suppression of an input signal (2) is specified, in particular for a hearing aid (60), the input signal (2) being distributed over a main signal path (8) and an auxiliary signal path (10), the auxiliary signal path (10 ) the presence of a useful signal component (40) and, if present, a parameter (22) for a frequency (42) of the useful signal component (40) is estimated, and in the main signal path (8) noise (44) in the input signal (2) by at least one Filter (32) with an infinite impulse response is suppressed, whose frequency response (50) has a maximum amount (52) in the range of the parameter (22) estimated frequency (42) of the useful signal component (40) while determining a complex pole (30) of the transfer function is implemented, the estimated parameter (22) for the frequency (42) of the useful signal component (40) being used to determine the complex pole (30) of the transfer function en will.