Speech Intelligibility Evaluation via Loudness-Weighted Disturbance Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech quality assessment algorithms, such as POLQA and PESQ, fail to accurately evaluate the intelligibility of degraded speech signals, which is crucial for information transfer and differs from sound quality evaluation.
Innovation Solution
A method that samples reference and degraded speech signals into frames, forms frame pairs, and calculates a difference function, compensating for disturbances using a loudness-dependent weighting value to reflect human auditory perception, thereby deriving an overall quality parameter indicative of intelligibility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional speech quality assessment algorithms (PESQ, POLQA) are used, then sound quality evaluation is achieved, but intelligibility evaluation is inaccurate
Solution Approach 1:
The patent applies local quality by differentiating between sound quality and intelligibility as separate evaluation dimensions. It introduces a specific intelligibility metric that operates locally on the speech signal's information content rather than treating all quality aspects uniformly, thereby improving intelligibility evaluation accuracy without compromising overall assessment reliability
Solution Approach 2:
The patent segments the quality assessment process into distinct components: sound quality evaluation and intelligibility evaluation. By separating these functions and applying specialized algorithms for each, the system achieves more accurate intelligibility measurement while maintaining the reliability of comprehensive quality assessment
2Measurement precision
If loudness-dependent weighting is applied to disturbance functions, then intelligibility assessment accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent implements dynamic weighting where the disturbance function is weighted according to the loudness of the reference speech signal. This dynamic adjustment allows the algorithm to focus computational resources on time periods when speech is actually present and intelligibility matters, improving assessment accuracy while managing complexity through adaptive rather than static processing
Solution Approach 2:
The patent changes the weighting parameter of the disturbance function based on loudness levels. By adjusting this parameter dynamically according to signal characteristics, the system achieves more accurate intelligibility assessment without requiring fundamentally more complex algorithmic structures
Data Source
AI summary
The present invention relates to a method of evaluating intelligibility of a degraded speech signal received from an audio transmission system conveying a reference speech signal. The method comprises sampling said reference and degraded signals into reference and degraded signal frames, and forming frame pairs by associating reference and degraded signal frames with each other. For each frame pair a difference function representing disturbance is provided, which is then compensated for specific disturbance types for providing a disturbance density function. Based on the density function of a plurality of frame pairs, an overall quality parameter is determined. The method provides for weighing disturbances in silent periods dependent on the loudness of the reference signal.