Audio Noise Reduction via Sub-band Segmentation and Energy Thresholding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional noise reduction systems in automatic speech recognition (ASR) systems often degrade the accuracy of identifying words in audio signals by introducing processing artifacts or removing portions of the audio spectrum, leading to 'false positives' and incomplete speech recognition.
Innovation Solution
A noise reduction system that uses a filter bank to decompose audio signals into sub-band signals, analyzes signal energies, and adjusts noise thresholds to attenuate noise elements, thereby removing noise without degrading speech elements, utilizing filters compatible with other audio processing systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If conventional noise reduction systems are used to remove environmental noise from audio signals, then noise is reduced, but processing artifacts are introduced and speech recognition accuracy decreases
Solution Approach 1:
The audio signal is divided into multiple frequency bands using a filter bank, allowing independent processing of each band. This segmentation enables selective attenuation of noise in specific frequency ranges while preserving speech components, thereby reducing processing artifacts and maintaining speech recognition accuracy.
Solution Approach 2:
Different frequency bands are processed with different attenuation characteristics based on their local signal properties. The system analyzes the energy distribution across frequency bands and applies selective noise reduction only where necessary, preserving the quality of speech-containing bands while removing noise from noise-dominated bands.
2Object-affected harmful factors
If conventional noise reduction techniques are applied to audio signals, then noise is attenuated, but portions of the audio spectrum are removed
Solution Approach 1:
The audio spectrum is segmented into multiple frequency bands using a filter bank, allowing precise control over which portions of the spectrum are attenuated. This enables selective noise reduction in specific bands while preserving other portions of the audio spectrum that contain speech information.
Solution Approach 2:
The system dynamically adjusts the attenuation parameters for different frequency bands based on the analyzed energy distribution. By changing the processing parameters adaptively, the system removes noise from noise-dominated frequency regions while maintaining the integrity of speech-containing frequency regions.
3Object-affected harmful factors
If noise reduction processing is applied to audio signals, then noise elements are removed, but processing artifacts are introduced
Solution Approach 1:
By segmenting the audio signal into frequency bands, the system can apply noise reduction locally to each band without affecting others. This segmented approach prevents the introduction of processing artifacts in speech-containing bands while effectively removing noise elements from noise-dominated bands.
Solution Approach 2:
The system analyzes the energy distribution of frequency bands and uses this feedback to dynamically adjust the noise reduction processing. This feedback mechanism allows the system to adapt to changing signal conditions and avoid introducing processing artifacts by applying appropriate attenuation levels based on real-time signal characteristics.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments of the present disclosure set forth a method of decomposing an audio signal into a set of sub-band signals and detecting a set of signal energy values, where each signal energy value is associated with a sub-band signal. The method also includes generating a noise reduction threshold based on at least one sub-band signal, and, for each sub-band signal, comparing the associated signal energy value to the noise reduction threshold. Based on determining that at least one sub-band signal is associated with a signal energy value below the noise reduction threshold, the method includes attenuating the at least one the sub-band signal to generate a set of attenuated sub-band signals. The method also includes combining at least one sub-band signal included in the set of sub-band signals with at least one attenuated sub-band signal included in the set of attenuated sub-band signals to generate an output audio signal.