Voice Quality Scoring via Digital Signal Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for testing voice transmission quality in building intercom systems are inaccurate due to a lack of direct evaluation of the degree of restoration and incomplete testing of continuous frequencies, leading to inconsistent distortion measurements and subjective evaluation discrepancies.
Innovation Solution
A system and method that segment and analyze continuous audio signals using a digital signal flow, performing spectrum transformations and calculating contrast score values to determine the degree of restoration by comparing input and output voice signal segments, ensuring accurate assessment of sound output performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional voice frequency transmission quality testing is performed using complete set of sound testing with five technical parameters, then the testing process is standardized, but the accuracy and precision of the detecting result is insufficient due to lack of direct evaluation on degree of restoration
Solution Approach 1:
The patent replaces traditional mechanical/acoustic testing methods with digital signal processing. Specifically, it uses digital signal collection, spectrum transformation (FFT), and automated calculation of degree of restoration, substituting manual or indirect measurement methods with computational analysis to achieve higher precision in voice transmission quality evaluation
Solution Approach 2:
The patent introduces a CPU as an intermediary that collects digital signals from the output end, performs spectrum transformation, and calculates the degree of restoration. This intermediary processing unit bridges the gap between raw signal collection and final quality evaluation, enabling automated and precise measurement of voice transmission characteristics
2Measurement precision
If distortion testing uses single frequency point (200Hz, 400Hz) as in prior art, then the testing is simple, but the degree of restoration determination is inaccurate because it is inconsistent with practical multi tone point voice signals
Solution Approach 1:
The patent segments the voice signal into multiple frequency components through spectrum transformation (FFT). Instead of testing at single frequency points, the signal is divided into numerous frequency bins that represent the complete spectral content. This segmentation allows accurate measurement of distortion across all frequency components present in practical voice signals
Solution Approach 2:
The patent performs spectrum transformation as a preliminary action before distortion calculation. By transforming the time-domain signal into frequency-domain representation first, all frequency components are simultaneously analyzed, eliminating the need for sequential single-frequency testing and reducing overall testing time while improving accuracy
3Measurement precision
If all concerned continuous frequencies are not tested in prior art, then the testing process is simplified, but the testing result accuracy is reduced and creates larger difference between testing result and practical subjective evaluation
Solution Approach 1:
The patent replaces selective frequency testing with comprehensive continuous frequency analysis using Fast Fourier Transform (FFT). The digital signal processing system automatically analyzes all frequency components within the voice range (200-4000Hz and beyond), substituting manual frequency selection with automated spectral analysis that captures the complete frequency content of voice signals
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach improves the precision and accuracy of voice signal performance testing by directly evaluating the degree of restoration and accounting for continuous frequencies, providing a more objective and accurate evaluation of voice transmission quality.
Implementation Method 1
performing a spectrum transformation on each of the voice signal groups, so as to obtain a corresponding sound feature value after transforming each of the voice signal groups
Data Source
Figure 1
Figure 2~3
AI summary
A method and system for scoring human sound voice quality, the method comprising: a sound source part (110) generating a standard genuine human voice signal as an input voice signal, and inputting the input voice signal to the system under test (120) from the sound source part (110); transmitting the input voice signal in the system under test (120) and outputting the input voice signal from an output end of the system under test (120) as an output voice signal; collecting the continuous output voice signal; and segmenting and analyzing the signal flow of the collected output voice signal to determine the degree of restoration.