Echo Cancellation Circuit Using Adaptive Reference Signal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current echo reduction systems in hands-free devices, particularly in vehicles, face challenges with acoustic echo cancellation due to proximity between microphones and loudspeakers, non-linear phenomena, and background noise from distant speakers, leading to significant attenuation of the speech signal and instability in echo cancellation algorithms.
Innovation Solution
A digital processing circuit with an adaptive echo cancellation stage using a modified reference signal from the output of the amplifier, combined with a preliminary denoising stage upstream to reduce background noise and a selective noise reduction stage analyzing temporal coherence for improved echo suppression and noise reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If the loudspeaker power is increased to overcome ambient noise in vehicle environments, then the speech signal can be heard over background noise, but the acoustic echo level increases significantly (up to twenty times higher than speech signal)
Solution Approach 1:
The patent applies preliminary denoising to the reference signal before it is used in echo cancellation. By pre-processing the reference signal to remove background noise, the echo canceller can more accurately model and subtract the acoustic echo without being interfered by noise, thus allowing high loudspeaker power to be used without generating excessive echo
Solution Approach 2:
The patent implements a feedback mechanism where the processed speech signal from the near speaker is fed back through the loudspeaker and captured by the microphone to create an artificial reference signal. This feedback loop enables the system to adaptively model the acoustic path and dynamically cancel the acoustic echo in real-time, counteracting the echo generated by high-power loudspeaker operation
2Volume of moving object
If the distance between microphone and loudspeaker is reduced to create a compact autonomous device, then the device becomes more integrated and portable, but the acoustic echo becomes considerable (twenty times higher than speech signal)
Solution Approach 1:
By applying preliminary denoising to the reference signal, the system compensates for the close proximity between microphone and loudspeaker. The pre-processed reference signal enables more accurate echo modeling despite the short distance, allowing the compact autonomous design to function effectively without excessive echo
Solution Approach 2:
The patent replaces physical separation (mechanical solution) with signal processing (digital solution). Instead of increasing the distance between microphone and loudspeaker to reduce echo, the system uses digital signal processing techniques including preliminary denoising and adaptive filtering to cancel the echo electronically, maintaining the compact mechanical design
3Object-generated harmful factors
If echo cancellation is applied to suppress acoustic echo, then the echo level is reduced, but the useful speech signal from the near speaker is also attenuated
Solution Approach 1:
By pre-denosing the reference signal before echo cancellation, the system improves the quality of the reference input to the adaptive filter. This allows the echo canceller to more accurately distinguish between echo and speech components, reducing unwanted attenuation of the useful speech signal while still suppressing echo
Solution Approach 2:
The patent changes the parameters of the reference signal by applying preliminary denoising processing. This modification of the reference signal characteristics enables the echo cancellation algorithm to operate more effectively, adjusting its filtering parameters to better preserve speech while removing echo
4Loss of information
If background noise reduction is applied to preserve speech in noisy environments, then speech clarity is improved, but the processing complexity increases
Solution Approach 1:
The patent applies preliminary denoising to the reference signal as a pre-processing step before echo cancellation. This early noise reduction simplifies subsequent processing stages by providing a cleaner reference signal, reducing the computational burden on later echo cancellation and speech enhancement algorithms
Solution Approach 2:
The patent segments the noise reduction process into distinct stages: preliminary denoising of the reference signal, echo cancellation, and final speech enhancement. This segmentation allows each stage to focus on specific aspects of signal processing, improving overall efficiency and reducing the complexity of any single processing block
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Significantly reduces acoustic echo and background noise, enhancing speech clarity and stability of echo cancellation, even in noisy environments and double talk situations, by accurately modeling non-linear phenomena and adapting to changing noise conditions.
Implementation Method 1
acoustic echo, that is to say the reverberation of the sound reproduced from above -loudspeaker-
Implementation Method 2
the phenomenon of reverberation of the environment of the speaker
Data Source
Figure 1~2
AI summary
The device comprises: a circuit for gathering acoustic signals with a microphone (18); a circuit for restoring audio signals with an amplifier (30) and a loudspeaker (16); a circuit (20) for interfacing with a telephone network or a portable telephone; and a circuit for digital processing of audio signals, comprising means for reducing the acoustic echo resulting from the interaction of the loudspeaker with the microphone. These means comprise: an echo cancellation stage (40), with a linear adaptive filter able to subtract from the signal picked up by the microphone a reference signal derived from the signal received at the output of the interfacing means; a stage (42) for removing the residual echo with gain control; and a stage (44) for selectively reducing the background noise present in the signal received at the output of the echo removal stage. There is moreover provided a stage (50) for selectively reducing the background noise present upstream in the signal received at the output of the interfacing means for application to the circuit for restoring audio signals, which stage comprises means for analysing temporal coherence of the signal, which are able to determine an a priori probability of presence/absence of speech on the basis of the respective energy levels in the spectral domain, and to use this probability to estimate a noise spectrum and to derive a denoised estimate of the signal received.