Non-Intrusive Speech Quality Assessment Parameter
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current non-intrusive speech quality assessment systems rely on subjective Mean Opinion Scores (MOS) from human listeners, which are time-consuming and costly, and lack effective parameters for real-time quality measurement in telecommunications networks.
Innovation Solution
A method generating a parameter from voiced signal sequences using frequency transforms, pitch frequency estimation, and averaging differences between frequency portions, allowing for improved speech quality assessment without intrusive testing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human listeners are used to assess speech quality through subjective tests, then measurement precision is improved, but loss of time and cost increase
Solution Approach 1:
The patent replaces the mechanical system of human listeners performing subjective tests with an automated computer-based system. The system uses signal processing algorithms to extract acoustic features and predict MOS scores automatically, eliminating the need for human panelists while maintaining measurement precision.
Solution Approach 2:
The patent changes the parameters used for quality assessment from subjective human ratings to objective acoustic parameters such as spectral characteristics, temporal patterns, and pitch variations. By transforming speech signals into these measurable parameters and using them to predict MOS, the system achieves both automation and precision.
2Measurement precision
If intrusive testing methods are used to assess quality, then measurement precision is improved, but device complexity and operational disruption increase
Solution Approach 1:
The system performs self-service quality assessment by automatically analyzing speech signals without requiring external test equipment or human intervention. The automated algorithm processes the speech signal directly, extracting features and predicting quality metrics independently, thereby simplifying the overall system architecture.
Solution Approach 2:
The patent replaces complex manual testing procedures with automated signal processing mechanisms. Instead of requiring physical test setups and human evaluators, the system uses computational algorithms to automatically assess quality, reducing device complexity while maintaining precision.
3Ease of operation
If non-intrusive testing is implemented, then ease of operation is improved, but measurement precision may deteriorate
Solution Approach 1:
The patent changes from subjective human parameters to objective acoustic parameters that can be automatically extracted from speech signals. By using spectral features, temporal characteristics, and other measurable acoustic properties, the system achieves both ease of automated operation and maintained measurement precision.
Solution Approach 2:
The system incorporates feedback mechanisms where the predicted MOS scores are continuously refined based on the analysis of acoustic parameters. This feedback loop allows the automated system to improve its precision over time while maintaining ease of operation through continuous automatic adjustment.
Data Source
AI summary
This invention relates to a new parameter suitable for use in non-intrusive speech quality assessment system. The invention provides a method of generating a parameter from a signal comprising a sequence of values measured from voiced portions of said signal at a sampling frequency, said parameter suitable for use in a quality assessment tool. The method includes steps of selecting portions of frequency transformed sections of the signal in dependence upon a pitch estimate; generating an average value for each portion; and generating a section parameter depending upon the difference between the averages of successive portions. Said section parameter is averaged over a number of iterations of the method to generate the new parameter of the invention.