Far Field Noise Suppression Using Reference Microphone
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing telephony devices face challenges in effectively suppressing far field noise, particularly in noisy environments, due to limitations in microphone arrays and stationary noise cancellation techniques, which can lead to unsatisfactory listening experiences and increased speaker effort.
Innovation Solution
A noise suppression system that utilizes a main microphone and a reference microphone, with a digital signal processor (DSP) to estimate and subtract far field noise from the main microphone signal, allowing for dynamic microphone roles and flexible placement, and includes methods for aligning audio frames and detecting local talker activity to optimize noise suppression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If beamforming technique is used with a microphone array, then directional sensitivity and signal to noise ratio are improved, but device complexity increases due to requiring multiple microphones in fixed arrangements
Solution Approach 1:
The patent extracts the noise estimation function from the main signal path by using a separate reference microphone that captures only noise. This allows the main microphone to focus on capturing voice signals while the reference microphone handles noise characterization, simplifying the overall system architecture compared to full beamforming arrays.
Solution Approach 2:
The patent segments the audio signal into voice components and noise components by using two separate microphones with different functional assignments. The main microphone captures voice + noise, while the reference microphone captures only noise, allowing independent processing of each component.
2Reliability
If stationary noise cancellation technique is used, then noise suppression is achieved, but instantaneous noise sources cannot be captured and processing delay increases
Solution Approach 1:
The patent transitions from static noise modeling to dynamic noise tracking by continuously updating the noise estimate from the reference microphone at every audio frame. This allows the system to adapt to changing noise conditions in real-time without the delays associated with long-term averaging methods.
3Measurement precision
If reference microphone captures noise over long period, then stationary noise is effectively captured, but instantaneous noise sources are lost and voice signal may be removed
Solution Approach 1:
The patent implements feedback by continuously monitoring the main microphone signal for voice presence and using this information to adjust the noise estimation process. When voice is detected, the system knows to be cautious about aggressive noise subtraction, preventing voice loss while maintaining noise suppression.
Solution Approach 2:
The patent changes the parameter of noise estimation from long-term averaging to short-term adaptive estimation. By updating the noise model at every audio frame based on current conditions, the system achieves both accuracy for stationary noise and responsiveness for instantaneous noise sources.
4Reliability
If additional microphones are added for beamforming, then noise suppression improves, but cost and space requirements increase
Solution Approach 1:
The patent makes the reference microphone serve multiple functions: capturing noise for suppression, detecting noise levels for dynamic adjustment, and providing spatial information about noise sources. This multi-functionality reduces the need for additional specialized components.
Data Source
AI summary
Noise suppression systems and methods suppress far field noise in a microphone signal. A telephony system includes a main microphone and a reference microphone. In one example, the main microphone and the reference microphone can be located in the same device. In another example, the main microphone and the reference microphone can be located in two separate devices. A DSP can use the reference microphone signal to carry out suppression of far field noise in the main microphone signal. In one approach the DSP can determine an estimate of far field noise in the main microphone signal based on a noise estimate of the reference microphone signal and a reference and main microphone coupling estimate, and then subtract the far field noise estimate from the main microphone signal. Alternatively, the DSP can suppress the main microphone signal if it determines that a local talker is inactive.