In-Vehicle Speech Recognition Echo Cancellation via Magnitude Ratio Mixing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing in-vehicle speech recognition systems face challenges in effectively canceling out multiple sounds output from multiple speakers, which can interfere with speech recognition processes, especially when bandwidth is insufficient for transmitting all output signals.
Innovation Solution
An information processing device that acquires sounds from multiple speakers, calculates the magnitude ratios of these sounds, and uses these ratios to generate a reference signal for canceling out the sounds by mixing sound signals in accordance with their magnitudes, thereby emphasizing louder sounds and effectively canceling out multiple speaker outputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all output signals from multiple speakers are transmitted for echo cancellation, then cancellation accuracy is improved, but bandwidth requirements increase
Solution Approach 1:
The patent extracts only the essential information needed for echo cancellation by calculating magnitude proportions of speaker outputs and using only the corresponding reference signals, rather than transmitting all speaker output signals. This reduces bandwidth requirements while maintaining cancellation accuracy through selective signal processing.
Solution Approach 2:
The patent merges multiple speaker output signals into a single reference signal by mixing them according to their magnitude proportions. This consolidation reduces the number of separate signals that need to be transmitted, thereby reducing bandwidth consumption while preserving the echo cancellation capability.
2Measurement precision
If echo cancellation process is performed for multiple speakers, then speech recognition accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent changes the approach from processing individual speaker signals to processing a single reference signal that represents the combined speaker outputs. By transforming the problem into parameter space (magnitude proportions) rather than signal space, the processing complexity is reduced while maintaining speech recognition accuracy.
Data Source
AI summary
The control unit of the information processing device acquires a plurality of first sounds output from a plurality of speakers. The control unit of the information processing device calculates a ratio of magnitudes of the plurality of acquired first sounds. Then, the control unit sets the mixing ratio by applying the calculated ratio of the magnitudes of the plurality of first sounds as a mixing ratio that is a ratio at which the plurality of sound signals is mixed, and as a mixing ratio used to generate a reference signal for canceling the plurality of output sounds output from the plurality of speakers in response to the plurality of sound signals from the sound acquired from the microphone.