Echo Cancellation for Sound Source Azimuth Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional sound emission and collection devices face challenges in precisely selecting the voice of the main utterer due to echo interference, which increases processing load and makes it difficult to remove omnidirectional echoes effectively.
Innovation Solution
A sound emission and collection device with a first echo cancel section and multiple second echo cancel sections, using adaptive filters with simpler configurations to estimate the sound source's azimuth and remove echoes, while sharing filter coefficients to reduce processing load and facilitate quick adaptation to changing environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If echo removal is performed for all sound collection beam signals to precisely select the main utterer's voice, then selection precision is improved, but processing load increases significantly
Solution Approach 1:
The patent divides echo removal processing into two segments: (1) a first echo cancel section that performs full echo removal on the sound collection beam signal with the highest signal level, and (2) second echo cancel sections that perform simplified echo removal on other sound collection beam signals. This segmentation allows the system to achieve sufficient echo removal for azimuth estimation without the excessive processing load of removing echoes from all signals with full complexity.
Solution Approach 2:
The patent applies partial echo removal by using different echo cancellation approaches depending on the signal's rank. The highest signal level beam undergoes complete echo removal, while other beams undergo partial echo removal sufficient for azimuth estimation. This partial action approach achieves the necessary precision for selecting the main utterer without performing excessive echo removal on all signals.
2Device complexity
If simplified echo cancel sections are used for azimuth estimation, then processing load is reduced, but echo removal precision deteriorates
Solution Approach 1:
The patent applies local quality by using different echo cancellation configurations for different signals based on their purpose. The first echo cancel section uses a full configuration with more taps for the highest priority signal, while second echo cancel sections use simplified configurations with fewer taps for other signals. This local differentiation ensures that the simplified sections provide sufficient echo removal for azimuth estimation without the full processing burden.
3Measurement precision
If filter coefficients are independently estimated for each echo cancel section, then echo removal accuracy is improved, but processing time and complexity increase
Solution Approach 1:
The patent merges the filter coefficient estimation process by having the first filter coefficient estimation section share its estimated coefficients with the second filter coefficient estimation sections. The second sections initialize their filters using coefficients from the first section, thereby combining the estimation effort into a single primary estimation process rather than requiring independent estimation for each section. This significantly reduces processing time while maintaining sufficient accuracy.
Solution Approach 2:
The first filter coefficient estimation section performs preliminary coefficient estimation that serves as the foundation for other echo cancel sections. By establishing the filter coefficients in advance through the first section's estimation, the system prepares reusable coefficients that can be initialized in the second sections, avoiding redundant estimation work and reducing overall processing time.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Provided is a sound emission and collection device capable of estimating the azimuth of a sound source (such as a main utterer) precisely without any processing load. The sound emission and collection device (1) is connected with another sound emission and collection device via a network or the like. The sound emission and collection device (1) receives a sound signal from another sound emission and collection device, as a sound emission signal (FE), and emits the same from a speaker (SP). The sound emission and collection device (1) collects the sound at microphones (MIC1 to MIC3), and produces sound collection beam signals (NE1 to NE3) of different azimuths. The sound emission and collection device down-samples the individual sound collection beam signals (NE1 to NE3), and filters out the echoes of the down-sampled sound collection beam signals (DNE1 to DNE3). The sound emission and collection device selects the sound collection beam signal (DNE1') of the highest signal level from the echo-filtered sound collection beam signals (DNE1' to DNE3'). The sound emission and collection device filters out the echoes of a sound collection beam signal (NE1) from the sound collection azimuth (D1) of the sound collection beam signal (DNE1'), and transmits the same to another sound emission and collection device.