Cochlear Model Speech Quality Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for evaluating speech signal quality are time-consuming, expensive, and lack accuracy, particularly in quantifying multi-dimensional distortions, leading to inefficient evaluation and potential misprediction of Mean Opinion Score (MOS) in speech quality assessment.
Innovation Solution
The implementation of a physiological cochlear model that processes speech signals to extract place-based and time-based analysis vectors, allowing for the determination of foreground and background distortions, and subsequently estimating overall speech quality, providing a more accurate and objective measurement of speech quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If subjective human listening tests are used to evaluate speech signal quality, then measurement accuracy is improved, but time consumption and cost increase significantly
Solution Approach 1:
The patent creates an objective measurement system that copies and simulates human auditory perception through a cochlear model. Instead of using actual human listeners, the system uses computational models (cochlear model, auditory filter bank, feature extractors) to replicate human speech quality assessment, thereby maintaining measurement accuracy while eliminating time-consuming subjective tests
Solution Approach 2:
The patent replaces the mechanical human listening process with an automated computational system. The cochlear model processes speech signals through multiple analysis stages (temporal, spectral, cepstral features) to objectively quantify speech quality, substituting human ears and brains with electronic signal processing that achieves comparable or superior measurement precision without time loss
2Loss of time
If traditional objective methods are used to evaluate speech signal quality, then time consumption is reduced, but measurement accuracy and ability to quantify multi-dimensional distortions deteriorates
Solution Approach 1:
The patent segments speech quality assessment into multiple independent distortion dimensions (roughness, harshness, breathiness, etc.) using a comprehensive feature extraction system. Each dimension is measured separately through specific feature analyzers (temporal features for roughness, spectral features for harshness, etc.), enabling accurate multi-dimensional distortion quantification that traditional single-score objective methods cannot achieve
Solution Approach 2:
The patent transitions from traditional single-dimensional speech quality scoring to multi-dimensional distortion analysis. By adding temporal, spectral, and cepstral feature dimensions, the system provides a comprehensive characterization of speech quality that captures multiple aspects of distortion simultaneously, achieving both speed and accuracy
3Measurement precision
If comprehensive speech quality analysis is performed to identify multiple distortion dimensions, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent divides the complex speech quality analysis into modular functional blocks: cochlear model stage, auditory filter bank stage, feature extraction stage (temporal, spectral, cepstral), and distortion dimension analysis stage. Each module performs a specific function, making the overall complex system manageable and implementable through standardized signal processing components
Solution Approach 2:
The cochlear model and feature extraction system serve multiple functions simultaneously: they extract features for multiple distortion dimensions, provide temporal and spectral analysis, and generate inputs for various quality metrics. This multi-functionality reduces overall system complexity by sharing common processing infrastructure across different analysis tasks
Data Source
AI summary
A method for feature extraction by an electronic device is described. The method includes processing speech using a physiological cochlear model. The method also includes analyzing sections of an output of the physiological cochlear model. The method further includes extracting a place-based analysis vector and a time-based analysis vector for each section. The method additionally includes determining one or more features from each analysis vector.