Echo Suppression via Dynamic FIR Filter Power Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing echo cancellation techniques face challenges in accurately estimating echo power, leading to ineffective echo suppression, especially in scenarios with clock-drift, nonlinearities, and changes in the echo path, which can result in reduced call quality due to residual echo interference.
Innovation Solution
A method using a Finite Impulse Response (FIR) filter dynamically adapted in the time domain to estimate echo power, rather than the echo signal, for applying frequency-dependent echo suppression, which is more robust to disturbances and requires lower accuracy, thereby improving echo suppression in communication systems like VoIP.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If echo cancellation techniques are used to estimate and subtract echo signal, then echo suppression is achieved, but measurement precision of echo power estimation deteriorates due to clock-drift, nonlinearities, and echo path changes
Solution Approach 1:
The patent replaces the traditional mechanical approach of echo signal estimation and subtraction with a statistical power estimation method. Instead of attempting to accurately model and subtract the echo signal (which requires high precision and fails under clock-drift and nonlinearities), the invention estimates only the echo power using a power estimator that processes microphone and speaker signals through filtering and correlation operations. This substitution of signal-based estimation with power-based statistical estimation resolves the contradiction by achieving reliable echo suppression without requiring precise echo power measurement.
2Measurement precision
If accurate echo signal estimation is required for echo suppression, then echo power measurement precision improves, but device complexity increases due to the need for precise modeling of echo path
Solution Approach 1:
The patent extracts only the essential information needed for echo suppression - the echo power - rather than attempting to extract and process the complete echo signal. The power estimator selectively processes signals to obtain power measurements through filtering operations (low-pass filtering of microphone signal, high-pass filtering of speaker signal) and correlation computations. This extraction of only the necessary power information, discarding the complexity of full signal modeling, resolves the contradiction by achieving sufficient measurement precision without proportionally increasing device complexity.
3Reliability
If frequency-dependent echo suppression is applied to suppress echo, then echo power in received signal is reduced, but near-end speech signal quality may deteriorate due to inappropriate suppression levels
Solution Approach 1:
The patent applies local quality by implementing frequency-dependent echo suppression where different suppression levels are applied to different frequency bands. The power estimator processes signals through frequency-selective filtering (low-pass and high-pass filters) to estimate echo power in different frequency regions. The echo suppressor then applies tailored suppression gains to each frequency band based on the estimated echo power, rather than applying uniform suppression across all frequencies. This localized approach resolves the contradiction by suppressing echo effectively in frequencies where it dominates while preserving near-end speech quality in frequencies where speech is predominant.
Data Source
AI summary
Method, user device and computer program product for suppressing echo. An audio signal is output from a speaker. A microphone receives an audio signal, wherein the received audio signal includes an echo resulting from the outputted audio signal. A Finite Impulse Response filter estimate h(n) (h with ^ symbole above h) is dynamically adapted in the time domain based on the outputted audio signal and the received audio signal to model an echo path h(n) related to the echo in the received audio signal. The filter estimate h(n) (h with ^ symbol above h) is used in an estimate of the echo power in the received audio signal, and the estimated echo power is used to apply echo suppression to the received audio signal, thereby suppressing the echo in the received audio signal.