DJ Transform Spectrogram for Fundamental Frequency Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting fundamental frequencies, such as short-time Fourier Transform, face limitations in simultaneously increasing temporal and frequency resolution due to the Fourier uncertainty principle, resulting in low precision in identifying speaker characteristics like gender and age from voice signals.
Innovation Solution
A fundamental frequency extraction method based on DJ transform, which generates a spectrogram indicating estimated pure-tone amplitudes for natural frequencies and calculates degrees of suitability using moving averages or standard deviations, allowing for the extraction of fundamental frequencies through local maximum values, thereby improving resolution and precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If short-time Fourier Transform is used to extract fundamental frequency, then frequency resolution can be improved by using longer sound duration, but temporal resolution deteriorates
Solution Approach 1:
The patent applies segmentation by dividing the analysis into multiple stages: initial fundamental frequency estimation using short-time Fourier transform, followed by DJ transform analysis with multiple springs. Each spring segment analyzes specific frequency ranges, and the results are combined to achieve both high frequency and temporal resolution simultaneously.
Solution Approach 2:
The patent employs dynamic adjustment of analysis parameters based on the input sound characteristics. The system adaptively selects spring parameters, damping ratios, and analysis windows according to the detected fundamental frequency and signal properties, allowing optimal resolution in varying conditions.
2Loss of time
If short-time Fourier Transform is used to extract fundamental frequency, then temporal resolution can be improved by using shorter sound duration, but frequency resolution deteriorates
Solution Approach 1:
The patent segments the frequency analysis into multiple parallel spring systems, each with different natural frequencies and damping ratios. This allows simultaneous analysis of multiple frequency components with short time windows, achieving both high temporal and frequency resolution through parallel processing of segmented frequency bands.
Solution Approach 2:
The patent transitions from traditional time-frequency analysis to a multi-dimensional spring-based resonance system. By introducing the dimension of spring resonance frequencies and damping ratios, the system achieves resolution in both time and frequency domains simultaneously through the interaction of multiple resonant systems.
Data Source
AI summary
A method of extracting a fundamental frequency of an input sound includes generating a DJ transform spectrogram indicating estimated pure-tone amplitudes for respective natural frequencies of a plurality of springs and a plurality of time points by calculating the estimated pure-tone amplitudes for the respective natural frequencies by modeling an oscillation motion of the plurality of springs having different natural frequencies with respect to an input sound, calculating degrees of fundamental frequency suitability based on a moving average of the estimated pure-tone amplitudes or on a moving standard deviation of the estimated pure-tone amplitudes with respect to each natural frequency of the DJ transform spectrogram, and extracting a fundamental frequency based on local maximum values of the degrees of fundamental frequency suitability for the respective natural frequencies at each of the plurality of time points.