Multi-Channel Audio Processing for Acoustic Echo De-Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing acoustic echo cancellation techniques for multi-channel audio produce artefacts and fail to effectively suppress the spatiality of acoustic echoes, especially in low-latency communications where echoes can quickly become distracting due to changing perceived directionality.
Innovation Solution
The apparatus suppresses spatiality of acoustic echoes by de-localizing the sound source, spreading it across multiple audio channels, and using adaptive filtering and echo cancellation artefact masking to make the echo diffuse and directionless, thereby reducing its perceptual impact.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If acoustic echo cancellation is applied to captured multi-channel audio, then acoustic echo is reduced, but acoustic echo cancellation artefacts are produced and spatiality of the echo is not effectively suppressed
Solution Approach 1:
The audio signal is segmented into multiple spatial channels, with different processing applied to each channel. The acoustic echo cancellation is performed separately on each spatial channel, allowing selective suppression of echo spatiality while preserving speech spatial characteristics. This segmentation enables the system to treat echo and speech differently in the spatial domain.
Solution Approach 2:
Different quality processing is applied to different spatial components of the audio signal. The echo, which typically arrives from specific directions, has its spatiality suppressed through de-localization techniques. Meanwhile, the speech signal maintains its original spatial characteristics. This local quality approach ensures that only the harmful echo spatiality is modified while preserving the useful speech spatial information.
2Object-affected harmful factors
If traditional acoustic echo cancellation is used, then echo is reduced, but the spatiality of acoustic echoes is not suppressed making them distracting in low-latency communications
Solution Approach 1:
Instead of trying to completely eliminate the echo (traditional approach), the invention inverts the approach by deliberately de-localizing the echo spatiality. The echo is transformed from having a specific directional character (which is distracting) to having a diffuse, non-directional character (which is less distracting). This inversion of the spatial characteristics makes the echo less perceptually disturbing.
Solution Approach 2:
The problem is solved by operating in the spatial dimension rather than just the temporal or frequency dimension. By applying spatial processing techniques to de-localize the echo across multiple spatial channels, the invention transforms the echo's spatial distribution. This dimensional approach to echo suppression addresses the perceptual distraction issue by changing how the echo is positioned in the spatial audio field.
Data Source
AI summary
An apparatus comprising means for:rendering a received output audio signal via at least one audio output;locally capturing multi-channel input audio, wherein the multi-channel input audio comprises multiple audio input channels comprising at least:a first audio input channel comprising audio captured at a first microphone; anda second audio input channel comprising audio captured at a second microphone; andprocessing the multi-channel input audio to produce a multi-channel input audio signal representing local captured audio, wherein processing the multi-channel input audio comprises suppressing spatiality of an acoustic echo wherein suppressing the spatiality of an acoustic echo comprises de-localizing a sound source representing the acoustic echo