Acoustic Phase Demodulation for High-Resolution Frequency Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current frequency analysis techniques, such as Fourier transforms, suffer from limited resolution due to the time-frequency uncertainty principle, making it difficult to accurately estimate partial frequencies and pitches in acoustic signals, especially in non-perfectly periodic sounds, which limits their application in sound analysis.
Innovation Solution
A method involving demodulating the phase of acoustic signals in the frequency domain, calculating a constancy index using a low-pass filter and complex exponential conversion, and parameterizing it with adjustable frequency precision to enhance frequency resolution, allowing for finer analysis of acoustic data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If Fourier transform is used for frequency analysis, then the method is simple and widely applicable, but the frequency resolution is limited due to time-frequency uncertainty principle
Solution Approach 1:
The patent segments the frequency analysis process into multiple discrete frequency points. Instead of treating the spectrum as a continuous function, the method analyzes each frequency point independently through phase demodulation, allowing high resolution at each point while maintaining overall computational feasibility.
Solution Approach 2:
The patent introduces phase demodulation as an intermediary step between the Fourier transform and final frequency measurement. By demodulating the phase at each frequency point and measuring its constancy, the system achieves high frequency resolution without requiring excessively long analysis windows that would compromise time resolution.
2Measurement precision
If analysis window duration is increased to improve frequency resolution, then frequency precision improves, but time resolution deteriorates
Solution Approach 1:
The patent employs dynamic phase demodulation that adapts to the local characteristics of the signal at each frequency point. By measuring phase constancy dynamically across multiple short-time windows and using low-pass filtering to detect phase variations, the system achieves high frequency precision without requiring a single long analysis window, thus preserving time resolution.
Solution Approach 2:
The patent changes the parameter being measured from raw spectral amplitude to phase constancy. By demodulating the phase and measuring its stability over time, the system can achieve high frequency resolution through phase coherence measurement rather than amplitude concentration, allowing shorter analysis windows and better time resolution.
3Adaptability or versatility
If traditional spectral analysis is applied to non-perfectly periodic signals, then the analysis can be performed, but accurate pitch estimation becomes difficult due to inharmonicity
Solution Approach 1:
The patent uses feedback through phase constancy measurement. By continuously monitoring the phase at each frequency point across multiple time windows and measuring how constant the phase remains, the system can identify the true fundamental frequency even in the presence of inharmonicity. The phase constancy metric provides feedback that distinguishes periodic components from aperiodic noise.
Solution Approach 2:
The patent replaces the traditional mechanical approach of peak detection in the amplitude spectrum with a phase-based measurement system. Instead of relying on amplitude concentration that fails for inharmonic signals, the system uses phase demodulation and constancy measurement, which remains effective for detecting periodicity even when the spectrum is not perfectly harmonic.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
The invention relates to a method for processing acoustic data, comprising steps of: receiving (110) acoustic data corresponding to an acoustic signal transformed in the frequency domain from an acoustic signal in the time domain, for example comprising a vibrato, or a beat, or a glissando, recorded by at least one acoustic sensor for at least one frequency of the frequency domain; demodulation (118) of the phase of the transformed acoustic signal; and calculation (120) of a stability coefficient of the demodulated phase; calculation (130) of a second stability coefficient of the demodulated phase on the basis of a demodulated phase difference between the instants of two successive frames; and display of said results, for example in the form of a helicoidal sonogram of the responses of the coefficients to a frequency vector. The invention can be applied to a tuning system or a music analysis system.