Speakerphone Feedback Attenuation via Dual Microphone Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for minimizing acoustic feedback in speakerphones, such as acoustic echo canceling, automatic microphone mixers, and directional microphones, have limitations that result in inadequate cancellation or practicality issues, particularly during changes in the acoustic environment or when multiple microphones are used.
Innovation Solution
A method using a speakerphone with two microphones that undergoes a calibration stage to match their frequency and phase responses, applying filter coefficients during a call to attenuate loudspeaker sounds while allowing user voices from the far field to be heard, effectively reducing acoustic coupling without disrupting user voices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If acoustic echo canceling is used to minimize acoustic coupling, then the adaptive filter can model and cancel echo under stable conditions, but the system fails to adapt quickly when acoustic signature changes occur during conversation
Solution Approach 1:
The system performs preliminary calibration by playing calibration sounds from the loudspeaker and recording them with microphones before actual use. This pre-characterization of the acoustic path allows the system to have filter coefficients ready before conversation begins, enabling immediate adaptation without waiting for far-end speech to trigger adaptive filtering.
Solution Approach 2:
The system separates calibration (performed when acoustic environment is stable) from operation (during conversation). The calibration stage captures the acoustic signature, and this information is stored for use during runtime, allowing the system to dynamically switch between adaptation and cancellation modes based on operational conditions.
2Reliability
If automatic microphone mixers are used to minimize loudspeaker-to-microphone coupling, then the audio feed to far end is cleaned up, but the teleconferencing system loses full-duplex functionality
Solution Approach 1:
The system introduces an acoustic model (filter coefficients derived from calibration) as an intermediary between the loudspeaker output and microphone input. This model allows the system to mathematically cancel echo paths while maintaining independent microphone and loudspeaker channels, preserving full-duplex operation unlike automatic microphone mixers that require muting.
3Reliability
If directional microphones are used to reduce acoustic coupling, then loudspeaker sounds are blocked from being sensed, but sounds from call participants behind the loudspeaker are also blocked
Solution Approach 1:
The system replaces the mechanical directional sensitivity approach with a signal processing solution. Instead of relying on microphone physical orientation to block loudspeaker sounds, the system uses acoustic echo canceling with filter coefficients to mathematically remove loudspeaker contributions from microphone signals, preserving omnidirectional pickup characteristics.
4Reliability
If filter coefficients are applied during runtime to attenuate loudspeaker sounds, then acoustic feedback is reduced, but the system must distinguish between loudspeaker sounds and user voices
Solution Approach 1:
The system creates a copy of the loudspeaker signal (available from the audio processing chain) and uses it to generate expected microphone signals through the acoustic model. By comparing actual microphone signals with these predicted loudspeaker-induced signals, the system can accurately identify and cancel only the loudspeaker contributions without affecting user voice signals.
Data Source
AI summary
A method is disclosed for acoustic feedback attenuation at a telecommunications terminal. A speakerphone equipped with a loudspeaker and two microphones is featured. Signals from the two microphones are subjected to a calibration stage and then to a runtime stage. The purpose of the calibration stage is to match the microphones to each other by advantageously using both magnitude and phase equalization across the frequency spectrum of the microphones. During the runtime stage, the microphones monitor the ambient sounds received from sound sources, such as the speakerphone's users and the loudspeaker itself, during a conference call. The speakerphone applies the generated set of filter coefficients to the optimized microphone's signals. By combining the signal from the reference microphone with the filtered signal from the optimized microphone, the speakerphone is able to attenuate the sounds from the loudspeaker that would otherwise be transmitted back to other conference call participants.