Low-Delay Audio Echo Canceller Using Sub-Band FIR Synthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional audio echo cancellers face challenges with slow adaptation to changes in the acoustic environment and high processing power requirements in full band systems, while sub-band systems introduce algorithmic delay, which is undesirable in some applications.
Innovation Solution
An audio echo canceller system that divides the far-end audio signal and microphone signal into sub-bands, using sub-band FIR filters with adaptive tap updates and a FIR model synthesize filter to generate a full-band echo estimate without introducing algorithmic delay, thereby combining the benefits of both full band and sub-band processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If sub-band processing is used for echo cancellation, then processing power requirements are reduced and adaptation speed is improved, but algorithmic delay is introduced
Solution Approach 1:
The patent divides the far-end audio signal and microphone signal into multiple sub-bands using analyze filters. Each sub-band is processed independently by sub-band FIR filters with adaptive tap updates, reducing the processing power required per band while maintaining overall echo cancellation effectiveness. This segmentation allows parallel processing that improves adaptation speed without introducing significant algorithmic delay.
2Loss of time
If full band adaptive FIR filtering is used for echo cancellation, then algorithmic delay is avoided, but processing power requirements become exhaustive and adaptation speed is slow
Solution Approach 1:
Instead of processing the entire frequency band with a single full-band adaptive FIR filter, the patent segments the signal into multiple sub-bands. Each sub-band uses a separate adaptive FIR filter with fewer taps, reducing the total computational load while maintaining the zero algorithmic delay property by avoiding synthesis filter delays.
Solution Approach 2:
The patent transforms the problem from full-band time-domain processing to multi-band frequency-domain processing. By applying Fourier transforms to convert signals into sub-bands, the system achieves efficient parallel processing that reduces computational complexity while maintaining real-time performance without algorithmic delay.
3Manufacturing precision
If the FIR filter length is increased to improve echo cancellation accuracy, then manufacturing precision of echo estimation is improved, but processing power requirements increase
Solution Approach 1:
The patent divides the long FIR filter into multiple shorter sub-band FIR filters. Each sub-band filter requires fewer taps to achieve the same effective echo cancellation accuracy, reducing the computational complexity from O(N) to O(N/M) where M is the number of sub-bands, while maintaining overall precision through parallel processing of all sub-bands.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The invention discloses an audio echo canceller adjusted to provide an echo attenuated output signal from an echo added microphone signal consisting of a near-end audio signal and a far -end audio signal originated echo signal by adding an inverted echo estimate, generated by filtering the far-end audio signal with a full band PIR filter including a set of full band filter taps, which taps are generated from sub-band filter taps .