Skip to content

Conversation

@tedmoore
Copy link
Member

I think it would be great to have one more example for the BufMFCC SC help file. If anyone can think of an interesting thing to do for it, that would be great!

@tedmoore tedmoore self-assigned this Feb 18, 2022
@jamesb93 jamesb93 added the documentation Improvements or additions to documentation label Mar 4, 2022
Copy link
Member

@tremblap tremblap left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nice text again very thorough.

I'd remove trivia (("cepstral" is pronounced like "kepstral"))

I'd add the data compression explanation:
"This analysis is often used for timbral description and timbral comparison, compressing in a small number of comparable coefficients the overall contour of the spectrogram." then I would change paragraph, and keep the more mathy one that starts after (mel->dct)

there is a coding error : "just as in :fluid-obj:MelBands. numCoeffs coefficients" is what one reads in the rendered text

the mfcc0 paragraph could be streamlined to:
"Other than the 0th coefficient, MFCCs are unchanged by differences in the overall energy of the spectrum. This means that timbres that have similar spectral contour at different volumes will be identical, other than their MFCC0 value. This is useful to compare timbre: we can ignore MFCC0 by setting the parameter startCoeff to 1." (or something in better english but with less repetition and stressing the musicianly use of dismissing mfcc0)

@tedmoore
Copy link
Member Author

I'd remove trivia (("cepstral" is pronounced like "kepstral"))

I think this is actually quite important as the "correct" pronunciation of words is sometimes a weird indication of being "in the know" as a sort of gate-keeping indication. I remember the first time I heard someone pronounce "kepstral" after I had been saying "septral" and I thought to myself "oh, am I doing it wrong? 😬 ". I think just being very clear is more inclusive.

I'd add the data compression explanation:
"This analysis is often used for timbral description and timbral comparison, compressing in a small number of comparable coefficients the overall contour of the spectrogram." then I would change paragraph, and keep the more mathy one that starts after (mel->dct)

👍

the mfcc0 paragraph could be streamlined to:
"Other than the 0th coefficient, MFCCs are unchanged by differences in the overall energy of the spectrum. This means that timbres that have similar spectral contour at different volumes will be identical, other than their MFCC0 value. This is useful to compare timbre: we can ignore MFCC0 by setting the parameter startCoeff to 1." (or something in better english but with less repetition and stressing the musicianly use of dismissing mfcc0)

👍

@tedmoore tedmoore merged commit ed93c51 into dev Mar 24, 2022
@tedmoore tedmoore deleted the MFCC-helpfile-&-sc-example branch March 24, 2022 11:22
@tremblap
Copy link
Member

I think just being very clear is more inclusive.
indeed

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

documentation Improvements or additions to documentation

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants