Signal Processor Beam Selection for Speech Leakage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing signal processors face challenges with speech leakage and insufficient beam accuracy, leading to suboptimal speech-to-interference ratio and robustness issues in dynamic environments, particularly in scenarios with unknown and changing speech directions.
Innovation Solution
A signal processor with multiple beamforming modules, speech-leakage-estimation modules, and a beam-selection module that dynamically selects the best beamformer output based on speech-leakage estimation signals, minimizing speech leakage and adapting to changing speech directions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple beamforming modules are used to cover different directions, then the coverage and adaptability are improved, but speech leakage occurs between the different beams
Solution Approach 1:
A beam selection module is introduced as an intermediary component that receives beamformer output signals from multiple beamforming modules and selects the appropriate output based on speech leakage estimation. This mediator prevents speech leakage by choosing the optimal beam output rather than simply combining all beam outputs, thereby resolving the contradiction between multi-directional coverage and speech leakage prevention
Solution Approach 2:
The system implements feedback through speech leakage estimation modules that continuously evaluate the output signals from multiple beamforming modules. The beam selection module uses this feedback information to dynamically select the best beam output, ensuring that speech leakage is minimized while maintaining adaptability across different directions
2Device complexity
If beamforming modules focus into fixed angular directions, then the system complexity is reduced, but the tracking speed of changing speech directions is insufficient
Solution Approach 1:
The system achieves dynamic adaptability through the beam selection module, which dynamically selects among fixed beamforming modules based on real-time speech leakage estimation. This allows the system to track changing speech directions quickly by switching between pre-configured fixed beams, resolving the contradiction between simple fixed beam configuration and fast tracking capability
Solution Approach 2:
Multiple beamforming modules are pre-configured to cover different angular directions before operation. This preliminary setup allows the system to rapidly respond to direction changes by simply selecting from the pre-prepared beam configurations, achieving fast tracking without the complexity of dynamically adjusting beam parameters in real-time
3Reliability
If speech leakage estimation is performed using similarity measures between speech-reference-signal and noise-reference-signal, then the speech-to-interference ratio is improved, but the computational complexity increases
Solution Approach 1:
The system replaces complex mechanical or hardware-based speech separation methods with computational similarity measures. By using signal processing algorithms to estimate speech leakage based on similarity between reference signals, the system achieves improved speech-to-interference ratio through software-based computation rather than complex hardware configurations
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A signal processor comprising a plurality of microphone-terminals configured to receive a respective plurality of microphone-signals. A plurality of beamforming-modules, each respective beamforming-module configured to receive and process input-signalling representative of some or all of the plurality of microphone-signals to provide a respective speech-reference-signal, a respective noise-reference-signal, and a beamformer output signal based on focusing a beam into a respective angular direction. A beam-selection-module comprising a plurality of speech-leakage-estimation-modules, each respective speech-leakage-estimation-module configured to receive the speech-reference-signal and the noise-reference-signal from a respective one of the plurality of beamforming-modules; and provide a respective speech-leakage-estimation-signal based on a similarity measure of the received speech-reference-signal with respect to the received noise-reference-signal. The beam-selection-module further comprises a beam-selection-controller configured to provide a control-signal based on the speech-leakage-estimation-signals.