Multi-Microphone Beamforming Using Dominant Mic and Phase Coherence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing beamformers used in electronic devices to separate desired audio signals from interfering signals are computationally intensive, especially when finer azimuth and elevation angles are used, leading to incorrect detection and distorted output signals.
Innovation Solution
The electronic device employs a beamformer functional block that detects a dominant microphone based on entropy and phase coherence, generates a combined audio signal, and uses frequency normalized least mean square (FNLMS) filters with two sets of coefficients to filter out interfering signals, reducing computational effort and improving signal quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If beamformers scan peaks of cross power spectral density (CPSD) for audio signals received in various azimuth and elevation angles to determine direction of desired audio signals, then the direction detection accuracy is improved, but the computational complexity increases significantly
Solution Approach 1:
The patent segments the audio signal processing into distinct functional blocks: a beamformer functional block that performs directional filtering, a dominant microphone detector functional block that identifies the primary sound source direction, and an interference detector functional block that separates interfering signals. This segmentation allows each block to perform a specific function with optimized computational requirements, avoiding the need for exhaustive CPSD scanning across all azimuth and elevation angles.
Solution Approach 2:
The patent performs preliminary detection of the dominant microphone and interfering audio signals before final beamforming processing. By detecting the dominant microphone first and identifying interfering signals in advance, the system prepares filter configurations proactively, reducing the computational burden during real-time audio processing and avoiding the need for complex on-the-fly CPSD calculations.
2Device complexity
If coarser azimuth and elevation angles are used to reduce computational work, then the computational complexity is reduced, but the beamformer incorrectly detects the direction of desired audio signals resulting in distorted output
Solution Approach 1:
The patent replaces the traditional mechanical scanning approach of CPSD peak detection with a signal processing-based approach using entropy calculation and phase coherence analysis. Instead of mechanically scanning through multiple azimuth and elevation angles to find CPSD peaks, the system uses mathematical transformations (FFT, phase coherence measurement) to directly determine the dominant direction and interfering signal directions, achieving accurate detection with reduced computational complexity.
Solution Approach 2:
The patent changes the parameters used for direction detection from spatial grid scanning (azimuth and elevation angles) to signal domain parameters (entropy, phase coherence, frequency spectrum). By transforming the detection problem from spatial parameter space to signal parameter space, the system achieves accurate direction detection without requiring fine angular resolution, thereby reducing computational complexity while maintaining reliability.
3Measurement precision
If beamformers use finer azimuth and elevation angles for closely-spaced microphone arrays, then the direction detection precision is improved, but the computational intensity increases significantly
Solution Approach 1:
The patent extracts and processes only the most relevant signal characteristics for direction detection: the dominant microphone signal and interfering signal components. By extracting these specific components through entropy-based dominant microphone detection and phase coherence analysis, the system avoids the need to process all possible spatial directions with fine angular resolution, significantly reducing computational energy consumption while maintaining detection precision.
Solution Approach 2:
The patent performs partial action by focusing computational resources on detecting the dominant microphone and primary interfering signals rather than exhaustively analyzing all possible directions. The system applies entropy calculation and phase coherence measurement to identify the most significant signal sources, performing sufficient (but not excessive) analysis to achieve accurate direction detection without the computational burden of fine-grained full-sphere scanning.
Data Source
AI summary
An electronic device includes a plurality of microphones, each pair of microphones in the plurality of microphones being a respective distance from one another. The electronic device also includes a beamformer functional block that receives audio signals from each of the microphones. The beamformer functional block detects a dominant microphone from among the plurality of microphones based on the audio signals from each of the microphones and the distances between the microphones, the dominant microphone being in a closest direction to a source of desired audio. The beamformer functional block also detects interfering audio signals based on phase coherence between audio signals from the dominant microphone and audio signals from other microphones in the plurality of microphones. The beamformer functional block generates a beamformed audio output signal based on the audio signals and the interfering audio signals from each of the microphones.