Signal Processing Device Speech Enhancement Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Single channel speech enhancement technologies using a single microphone often result in lower speech recognition performance compared to noisy observation signals, and the effect on improving speech recognition is limited.
Innovation Solution
A signal processing device that includes a speech enhancement unit to generate an enhancement signal, an addition unit to add the observation signal to the enhancement signal, and a speech recognition unit to perform speech recognition on the modified enhancement signal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If single channel speech enhancement technology is used to remove noise from observation signal, then noise is reduced, but speech recognition performance becomes lower than using the noisy observation signal directly
Solution Approach 1:
The patent merges the noisy observation signal with the enhanced signal by adding them together. The addition unit combines these two signals to create a modified enhanced signal that preserves the speech quality improvements from enhancement while retaining the speech recognition performance characteristics of the original observation signal, thereby resolving the contradiction between noise reduction and recognition accuracy
Solution Approach 2:
The patent creates a composite signal by combining the observation signal and enhancement signal. This composite approach allows the system to benefit from both the noise reduction properties of speech enhancement and the speech recognition performance of the original signal, effectively creating a hybrid signal with superior overall performance
2Measurement precision
If multi-channel speech enhancement technology (beamformer) is used, then speech recognition performance is greatly improved, but device complexity increases due to requiring multiple microphones
Solution Approach 1:
The patent makes the single-microphone system perform multiple functions: it both enhances speech quality by removing noise and maintains speech recognition performance through signal addition. This multi-functional approach allows a simple single-microphone device to achieve results previously only attainable with complex multi-channel systems
Solution Approach 2:
The patent changes the processing approach rather than the hardware configuration. By modifying the signal processing parameters and methods (enhancing then adding the original signal), the system achieves multi-channel-like performance improvements while maintaining single-channel hardware simplicity
Data Source
AI summary
A signal processing device (10) includes: a speech enhancement unit (11) that generates, from an observation signal, an enhancement signal in which a voice of a speaker is enhanced; an original sound addition unit (12) that adds the observation signal to the enhancement signal; and a speech recognition unit (13) that performs speech recognition on the enhancement signal to which the observation signal is added by the original sound addition unit (12).