Vehicle Speech Signal Interference Reduction via Sparse Segmented Beamforming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional adaptive beamformers for speech signal processing in vehicles suffer from target signal cancellation due to steering-vector errors and reverberation, leading to interference and noise leakage, especially in complex acoustic environments like vehicles, where the far-field assumption is no longer valid.
Innovation Solution
A multi-stage system employing overlapping speech detection, blind speech separation, sparse generalized sidelobe cancellation, and post-processing techniques to distinguish and suppress noise and interference in different audio segments, using sparse regularization and log-spectral amplitude estimation to improve signal processing in vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If conventional adaptive beamforming with GSC is applied to cancel interference and noise, then interference and noise reduction is improved, but target signal cancellation occurs due to steering-vector errors
Solution Approach 1:
The patent segments the speech signal into different types (noise-only segments, single-speaker segments, overlapping speech segments) and applies different processing approaches to each segment. This segmentation allows the system to apply GSC only when necessary and avoid target signal cancellation in segments where it is not needed.
Solution Approach 2:
The patent dynamically adjusts the beamforming approach based on the observed data characteristics. By detecting the type of segment currently being processed, the system dynamically switches between different processing modes (noise suppression, single-speaker enhancement, overlapping speech handling), preventing unnecessary target signal cancellation.
2Object-affected harmful factors
If GSC is applied uniformly to entire utterance, then interference cancellation is improved, but processing efficiency deteriorates
Solution Approach 1:
The utterance is segmented into different types based on speech activity detection and overlap detection. Noise-only segments and single-speaker segments are identified and processed differently from overlapping speech segments, avoiding unnecessary GSC computation in segments where it is not needed.
Solution Approach 2:
Instead of applying GSC uniformly to the entire utterance, the patent applies GSC selectively only to overlapping speech segments where interference cancellation is actually needed. This partial action approach maintains interference cancellation performance where required while significantly improving processing efficiency.
3Device complexity
If far-field assumption is used in beamforming, then system complexity is reduced, but accuracy deteriorates in vehicle environments
Solution Approach 1:
The patent changes the fundamental parameter assumption from far-field to near-field modeling. By using near-field beamforming equations that account for spherical wave propagation and position-dependent amplitudes, the system achieves accurate signal processing in vehicle environments without excessive complexity.
Data Source
AI summary
Interference in an audio signal is reduced by estimating a target signal using beam-forming in a direction of the signal source. A set of estimates of interference is determined by using a microphone array filtering matrix to block the target signal in the audio signal. A set of filters is optimized by minimizing an objective function measuring a mismatch between the set of estimates of interference and the estimate of the target signal. The minimizing uses a sparse regularization of coefficients of the set of filters. The set of estimates of interference are filtered using the set of filters after the optimizing. Then, the estimate of interference after the optimizing is subtracted from the target signal.