Multi-microphone Source Tracking via Acoustic Fingerprinting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional multi-microphone source tracking and noise suppression methods fail to accurately differentiate between desired and interfering sources in acoustic environments with multiple sources behaving similarly, leading to poor speech signal quality and noise suppression.
Innovation Solution
The implementation of a system that uses steered null error phase transform (SNE-PHAT) for time delay of arrival estimation, adaptive blocking matrices, and switched super-directive beamforming to track and suppress noise, employing Gaussian mixture models for acoustic scene modeling and source identification to enhance speech signal clarity and reduce noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional ASA methods utilize spatial information such as TDOA or energy levels to locate acoustic sources, then the DS location can be estimated, but multiple acoustic sources may be present which behave similarly to the expected signature, making it impossible to accurately differentiate the DS from interfering sources
Solution Approach 1:
The patent introduces an acoustic fingerprinting intermediary that captures unique spectral characteristics of the desired source. This fingerprint acts as a mediator between the spatial location data and source identification, enabling differentiation of sources that occupy similar spatial positions but have distinct acoustic signatures. The fingerprint template matching process compares real-time spectral features against stored templates to reliably identify the desired source amidst interfering sources with similar spatial behavior.
2Manufacturing precision
If beamforming is steered to the DS based on angle of incidence, then sound from the DS can be better captured, but accurate source identification is required to avoid steering toward interfering sources
Solution Approach 1:
The acoustic fingerprint serves as an intermediary verification layer between source detection and beamforming steering. Before steering the beamformer to a detected source direction, the system queries the fingerprint database to confirm the source is the desired one. This prevents erroneous steering toward interfering sources that may exhibit similar spatial characteristics, ensuring beamforming accuracy is applied only to correctly identified targets.
Solution Approach 2:
The system implements feedback by continuously monitoring the acoustic scene, comparing detected source fingerprints against stored templates, and adjusting beamforming steering directions based on identification results. When a source is identified as the desired source through fingerprint matching, the beamformer steers toward that direction; when mismatched or ambiguous, the system maintains current steering or searches for alternative sources, creating a closed-loop control system that adapts to changing acoustic environments.
Data Source
AI summary
Methods, systems, and apparatuses are described for improved multi-microphone source tracking and noise suppression. In multi-microphone devices and systems, frequency domain acoustic echo cancellation is performed on each microphone input, and microphone levels and sensitivity are normalized. Methods, systems, and apparatuses are also described for improved acoustic scene analysis and source tracking using steered null error transforms, on-line adaptive acoustic scene modeling, and speaker-dependent information. Switched super-directive beamforming reinforces desired audio sources and closed-form blocking matrices suppress desired audio sources based on spatial information derived from microphone pairings. Underlying statistics are tracked and used to updated filters and models. Automatic detection of single-user and multi-user scenarios, and single-channel suppression using spatial information, non-spatial information, and residual echo are also described.