Speech Signal Noise Reduction via Prototype Spectral Envelopes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hands-free communication systems in vehicles suffer from poor speech signal quality due to background noise, particularly at high speeds, leading to a low signal-to-noise ratio and artifacts like the gating effect, which are not adequately addressed by current noise reduction methods.
Innovation Solution
A signal processing system that uses prototype spectral envelopes and reference noise prototypes to enhance speech signals, accounting for subsequent codec processing to reduce noise effectively, thereby improving speech quality and suppressing artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If conventional noise reduction methods (Wiener filters or spectral subtraction) are applied to speech signals in hands-free telephony, then some noise suppression is achieved, but the intelligibility and quality are still not improved sufficiently when perturbations are strong, and severe artifacts known as the gating effect occur at transitions from speech activity to speech pauses
Solution Approach 1:
The patent applies preliminary action by pre-processing the speech signal with noise reduction algorithms before the signal undergoes codec encoding and transmission. The system performs noise estimation and spectral subtraction in advance, creating a cleaned speech signal that is then encoded and transmitted. This preliminary noise reduction prevents the gating effect artifacts from occurring during subsequent codec processing, thereby maintaining speech intelligibility even in noisy environments with strong perturbations.
2Loss of information
If strong noise reduction is applied to improve speech quality in noisy environments, then the signal-to-noise ratio improves, but severe artifacts known as the gating effect occur at transitions from speech activity to speech pauses
Solution Approach 1:
The patent implements feedback by continuously monitoring the speech signal characteristics and dynamically adjusting the noise reduction parameters. The system estimates the speech activity level and noise spectrum in real-time, and uses this feedback to adaptively control the spectral subtraction process. This feedback mechanism prevents over-subtraction during speech pauses, thereby reducing the gating effect artifacts while maintaining adequate noise suppression during active speech segments.
Data Source
AI summary
The present invention relates to a method for signal processing comprising the steps of providing a set of prototype spectral envelopes, providing a set of reference noise prototypes, wherein the reference noise prototypes are obtained from at least a sub-set of the provided set of prototype spectral envelopes, detecting a verbal utterance by at least one microphone to obtain a microphone signal, processing the microphone signal for noise reduction based on the provided reference noise prototypes to obtain an enhanced signal and encoding the enhanced signal based on the provided prototype spectral envelopes to obtain an encoded enhanced signal.