Logarithmic Spectrum Harmonicity Estimation for Audio Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating harmonicity in audio signals face instability due to large dynamic ranges in the linear amplitude domain, leading to inaccurate results and limited contribution from high-frequency components, which are crucial for audio classification and noise estimation.
Innovation Solution
Calculating the Subharmonic-to-Harmonic Ratio (SHR) in the logarithmic spectrum domain, where a log amplitude spectrum is derived, and difference spectra are generated by subtracting odd and even multiple frequency sums, providing a monotonically increasing measure of harmonicity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If harmonicity is measured in the linear amplitude domain, then the calculation can be performed directly, but the large dynamic range leads to numerical instability and inaccurate results
Solution Approach 1:
The patent transforms the measurement domain from linear amplitude to logarithmic spectrum domain. This parameter change converts multiplicative relationships into additive ones, eliminating numerical instability caused by large dynamic ranges while preserving calculation feasibility through log-amplitude spectral analysis.
2Ease of manufacture
If harmonicity is measured in the linear amplitude domain, then the calculation is straightforward, but high frequency components contribute less due to the dynamic range limitation
Solution Approach 1:
By changing from linear amplitude to logarithmic spectrum representation, the patent equalizes the contribution of different frequency components. The logarithmic transformation compresses the dynamic range, allowing high-frequency components to contribute more equally to the harmonicity measurement without being overwhelmed by low-frequency energy.
3Reliability
If an approximation is used to calculate subharmonic-to-harmonic ratio in the linear domain, then numerical issues are avoided, but the results become inaccurate
Solution Approach 1:
The patent resolves the trade-off between numerical stability and accuracy by performing SHR calculation in the logarithmic domain. Here, the division operation becomes subtraction of log-amplitudes, which is numerically stable and exact, eliminating the need for approximations while maintaining high precision in the resulting harmonicity measure.
Data Source
AI summary
Embodiments are described for harmonicity estimation, audio classification, pitch determination and noise estimation. Measuring harmonicity of an audio signal includes calculation a log amplitude spectrum of audio signal. A first spectrum is derived by calculating each component of the first spectrum as a sum of components of the log amplitude spectrum on frequencies. In linear frequency scale, the frequencies are odd multiples of the component's frequency of the first spectrum. A second spectrum is derived by calculating each component of the second spectrum as a sum of components of the log amplitude spectrum on frequencies. In linear frequency scale, the frequencies are even multiples of the component's frequency of the second spectrum. A difference spectrum is derived subtracting the first spectrum from the second spectrum. A measure of harmonicity is generated as a monotonically increasing function of the maximum component of the difference spectrum within predetermined frequency range.