Hi Marc,
I would like to submit to you the problem affecting the Standard Deviation relative to the frequency of the H / V main peak.
I verified that with versions 3.0 and later there is a different value for average - minimum and average - maximum both considering f0 from windows and f0 from temporary window that appears on the H / V plot results.
When processing data files with Geopsy releases before 3.0 this does not happen.
Is there an explanation for this discrepancy? Is this a false problem?
I am attaching a pdf related to the processing of the same curve with 2.10 and 3.4.2 releases where this problem is evident.
Best regards
Luigi
f0 standard deviation
f0 standard deviation
- Attachments
-
- Geopsy comparison.pdf
- (330.48 KiB) Downloaded 88 times
Re: f0 standard deviation
Hi Luigi,
I'm not surprised that some minor differences exist. Since 2.10, many details changed (sampling, smoothing, rounding of number of samples to basic prime numbers for FFTW speed-up, geometric average for amplitudes,...). For your example, this is about 2% difference for the estimation of f0 range. I'm unsure if I did a complete comparison between 2.5.0 and 3.x.x for H/V. For sure I did it for linear FK, and there are minor differences that can be fully explained by the minor modification cited here above.
Best regards,
Marc
I'm not surprised that some minor differences exist. Since 2.10, many details changed (sampling, smoothing, rounding of number of samples to basic prime numbers for FFTW speed-up, geometric average for amplitudes,...). For your example, this is about 2% difference for the estimation of f0 range. I'm unsure if I did a complete comparison between 2.5.0 and 3.x.x for H/V. For sure I did it for linear FK, and there are minor differences that can be fully explained by the minor modification cited here above.
Best regards,
Marc