Adaptive Noise Suppression for Virtual Meetings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current noise suppression methods, such as conventional digital signal processing (DSP) algorithms and beamforming techniques, are inadequate for effectively reducing non-stationary noise and reverberation in dynamic environments, particularly in conference calls and online meetings.
Innovation Solution
The use of intelligent microphone arrays configured in various patterns, associated with machine learning models and processing engines, to perform sound source localization, noise suppression, and speech enhancement, generating anti-noise signals to cancel out unwanted noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional DSP algorithms are used for noise suppression, then stationary noise can be suppressed, but non-stationary noise cannot be effectively suppressed
Solution Approach 1:
The system dynamically adapts to different noise types by continuously analyzing the noise characteristics and adjusting the suppression strategy in real-time. The machine learning model processes audio signals to identify whether noise is stationary or non-stationary and applies appropriate suppression techniques accordingly, making the system flexible and adaptive to varying noise conditions.
Solution Approach 2:
The system changes processing parameters based on noise characteristics. When non-stationary noise is detected, the system adjusts suppression parameters and algorithms to match the temporal and spectral properties of the noise, enabling effective suppression across different noise types rather than using fixed parameters.
2Measurement precision
If beamforming techniques are used, then speech isolation can be achieved, but reverberation and noise from the same direction cannot be suppressed
Solution Approach 1:
The system merges beamforming with additional noise suppression techniques to create a comprehensive audio processing pipeline. By combining spatial filtering with spectral subtraction and machine learning-based noise estimation, the system addresses multiple types of interference including reverberation and noise from the same direction that beamforming alone cannot handle.
Solution Approach 2:
The system introduces an intermediary noise estimation and suppression stage between beamforming and speech output. This intermediary processing layer analyzes the beamformed signal, estimates remaining noise and reverberation components, and applies additional suppression to eliminate harmful factors that pass through the beamforming stage.
3Reliability
If headphones are used for noise isolation, then audio quality can be improved, but user comfort deteriorates
Solution Approach 1:
The system extracts and removes noise components from the audio signal processing path rather than physically isolating the user from noise through headphones. By using signal processing to identify and eliminate noise sources, the system maintains open-ear listening comfort while achieving the audio quality benefits normally requiring noise-isolating headphones.
Data Source
AI summary
One example method includes performing sound quality operations. Microphone arrays are used to cancel background noise and to enhance speech. With arrays at each environment of each user participating in a call, a first microphone array can cancel or suppress background noise and a second array can generate enhanced speech for transmission to other users. Thus, for user, the audio signal output by the user's device includes an anti-noise signal to cancel background noise present in the user's environment and enhanced speech from other users.