Voice Quality Evaluation Using Background Noise Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for objective evaluation of voice quality in speech signals do not account for the type of background noise, leading to inaccuracies in perceived quality assessment.
Innovation Solution
A method that classifies background noises into predefined classes and evaluates voice quality based on these classifications, using audio parameters like time and frequency indicators, and estimates loudness to calculate a voice quality score, providing a more accurate assessment closer to subjective perception.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current objective evaluation methods are used, then evaluation simplicity is maintained, but measurement precision of voice quality deteriorates due to not accounting for background noise type
Solution Approach 1:
The evaluation method is segmented into distinct stages: background noise classification (categorizing noise types such as traffic, restaurant, or indoor noise), speech quality evaluation (assessing quality under each noise category), and result combination (integrating both assessments). This segmentation allows the system to account for noise type effects while maintaining a structured, manageable evaluation process that improves precision without overwhelming complexity
Solution Approach 2:
The patent introduces an intermediary classification system that acts as a mediator between the speech signal and the final quality evaluation. By first classifying the background noise into predefined categories and then evaluating speech quality specific to each category, the system captures the nuanced impact of different noise types on perceived quality, thereby improving measurement precision
2Measurement precision
If background noise classification is added to evaluation, then measurement precision improves, but device complexity increases due to additional processing steps
Solution Approach 1:
The system changes parameters by introducing noise classification categories as an additional dimension to the evaluation process. By defining specific noise types (e.g., traffic noise, restaurant noise, indoor noise) and creating separate quality evaluation models for each category, the system captures the differential impact of noise types on speech quality, thereby improving measurement precision through parameter differentiation
3Measurement precision
If subjective evaluation methods are used, then measurement precision of perceived quality is improved, but loss of time increases due to human tester involvement
Solution Approach 1:
The patent creates an objective evaluation system that copies or replicates the key characteristics of subjective human perception. By training objective evaluation models on subjective evaluation data and incorporating noise classification that mirrors human noise type recognition, the system achieves high correlation with perceived quality while eliminating the time-consuming human tester involvement, thereby improving productivity
Data Source
AI summary
A method and device are provided for the objective evaluation of voice quality of a speech signal. The device includes: a module for extracting a background noise signal, referred to as a noise signal, from the speech signal; a module for calculating the audio parameters of the noise signal; a module for classifying the background noise contained in the noise signal on the basis of the calculated audio parameters, according to a predefined set of background noise classes; and a module for evaluating the voice quality of the speech signal on the basis of at least the resulting classification relative to the background noise in the speech signal.