Echo Suppressor Frequency Domain Ratio Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing echo suppression methods in multi-channel audio systems face challenges such as low capability in following large changes between single-talk and double-talk states, leading to erroneous voice recognition, increased cost and apparatus size, and residual echo errors due to independent sound intensity changes across channels.
Innovation Solution
An echo suppressor that converts reference and observation sound signals into the frequency domain, calculates and compares their ratio and correlation to determine an estimation coefficient for echo level, and suppresses echoes based on this estimation, eliminating the need for separate channel processing and adaptation learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional echo suppression methods using NLMS algorithm are applied to multi-channel audio systems, then echo cancellation is achieved, but the system fails to accurately follow large changes between single-talk and double-talk states, leading to erroneous voice recognition
Solution Approach 1:
The patent changes the fundamental parameters of echo suppression by transitioning from time-domain adaptive filtering (NLMS) to frequency-domain spectral subtraction. This involves converting signals via FFT, computing spectral ratios and correlations, and applying echo estimates in the frequency domain before inverse transformation. This parameter change enables the system to rapidly adapt to state transitions while maintaining suppression accuracy.
Solution Approach 2:
The patent replaces the iterative adaptive filtering mechanism (mechanical system) with a direct spectral analysis approach. Instead of using adaptive filters that require multiple iterations to converge, the system directly computes spectral ratios and correlations in the frequency domain, substituting the iterative mechanical process with a direct computational method that responds instantly to state changes.
2Measurement precision
If multiple suppressing mechanisms are provided for each channel in a multi-channel audio system, then echo suppression is achieved, but the cost and apparatus size increase
Solution Approach 1:
The patent merges the echo suppression processing across all channels by converting multi-channel signals to frequency domain representations and processing them through a unified spectral analysis framework. Instead of maintaining separate suppressing mechanisms for each channel, the system combines channel information in the frequency domain and applies a unified echo estimation and suppression process, thereby reducing the number of independent processing units required.
Solution Approach 2:
The patent transitions from time-domain channel-by-channel processing to frequency-domain holistic processing. By applying FFT to convert time signals into frequency representations, the system operates in a different dimension where multi-channel information can be processed simultaneously through spectral ratios and correlations, eliminating the need for multiple separate time-domain filtering mechanisms.
3Measurement precision
If adaptation processing based on NLMS algorithm is used, then echo estimation is performed, but the processing delay increases and real-time performance deteriorates
Solution Approach 1:
The patent performs preliminary spectral analysis by pre-computing FFT transforms of reference and observation signals, and pre-calculating spectral ratios and correlations before echo suppression is actually needed. This preliminary processing in the frequency domain prepares the necessary spectral characteristics in advance, so that when echo suppression is required, the system can quickly apply the pre-computed spectral information without undergoing lengthy iterative adaptation processes.
Data Source
AI summary
An apparatus is provided for suppressing an echo signal included in a measured signal corresponding to a measured sound. In the apparatus, the measured signal and a reference signal in a time domain are transformed into a frequency domain, and calculated for obtaining each value of a ratio and a correlation between the measured signal and the reference signal in the frequency domain. With executing a comparison of the values of the ratio and the correlation, a coefficient is derived, where a product of the coefficient and the measured sound in the frequency domain gives an estimated value of the echo signal. The echo in the measured signal is suppressed with subtracting the estimation of the echo signal from the measured signal, respectively in the frequency domain.