Adaptive Echo Canceller Using Frequency-Dependent Delay Lines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing echo canceller technologies in mobile audio devices face challenges in maintaining audio quality due to acoustic coupling between speaker and microphone, particularly in closed rooms where long echoes occur, and fail to efficiently utilize processing resources across different frequency spectrums.
Innovation Solution
An adaptive filter unit with a processor that performs FFT transformations of audio signals, calculates frequency-dependent time delays, and uses a multidelay filter structure with circular delay lines to effectively filter echoes, allowing for large time delays in low frequencies and short delays in higher frequencies, optimizing resource use and adapting to various hearing devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a fixed time delay is used in echo canceller for all frequencies, then the structure is simple, but it cannot effectively handle different echo characteristics in different frequency spectrums
Solution Approach 1:
The patent applies local quality by implementing frequency-dependent time delays where different frequency components (FFT bins) have different delay values. Low frequencies use larger delays to account for longer room echoes, while high frequencies use smaller delays. This allows the echo canceller to adapt to the specific echo characteristics of each frequency band, improving cancellation performance without applying a uniform complex structure across all frequencies.
Solution Approach 2:
The patent segments the frequency spectrum into multiple FFT bins and processes each bin independently with its own delay line. This segmentation allows different time delays to be applied to different frequency components, enabling the system to handle diverse echo characteristics across the spectrum while maintaining a manageable structure through modular processing.
2Reliability
If large time delays are applied to all frequencies, then long echoes are effectively cancelled, but processing resources are wasted on frequencies that require shorter delays
Solution Approach 1:
The patent optimizes processing resource efficiency by applying locally appropriate time delays to each frequency bin. Low frequencies, which typically exhibit longer room echoes, receive larger delay values. High frequencies, which have shorter echo characteristics, receive smaller delay values. This localized optimization ensures that processing resources are not wasted applying excessively large delays to frequencies that require minimal delay, thereby improving overall processing efficiency while maintaining effective echo cancellation.
3Productivity
If frequency-dependent time delays are implemented, then resource utilization is optimized, but the device complexity increases
Solution Approach 1:
The patent manages device complexity by segmenting the frequency spectrum into discrete FFT bins and assigning a separate delay line to each bin. This modular segmentation allows the system to implement frequency-dependent delays in a structured manner, where each frequency component is processed independently. The segmentation approach makes the complex task of frequency-dependent processing more manageable and implementable through systematic organization of delay lines and memory structures.
Solution Approach 2:
The patent employs preliminary action by pre-calculating and storing the frequency-dependent time delay values for each FFT bin in memory before actual echo cancellation processing. This pre-computation of delay parameters allows the processing stage to simply retrieve and apply the appropriate delay values without performing complex real-time calculations, thereby optimizing processing efficiency while managing device complexity through advance preparation of delay characteristics.
Data Source
AI summary
The invention relates to an adaptive filter unit, in particular for being used as an echo canceller, comprising a first filter input, configured to receive a first electric audio signal, indicative of a first audio signal A(t), a second filter input, configured to receive a second electric audio signal, indicative of a second audio signal B(t), a processor and a filter output. The processor is configured to calculate and provide audio estimation data X(fn, A(t1, . . . , tM(fn))) in the frequency domain; to calculate a transformed second audio signal Y(fn, B(t)), formed by a transformation of the second audio signal B(t) into the frequency domain; and to calculate a filtered audio signal by subtracting delayed audio estimation data from the transformed second audio signal, wherein the delayed audio estimation data is provided by a memory unit of the adaptive filter unit, which is arranged to provide a data exchange with the processor, and wherein the delayed audio estimation data comprises a frequency dependent time delay compared to the transformed second audio signal.