Dear Members of this Forum,
I’m trying to convert the contimbre database in a orchidea-usable database.
Contimbre has a lot of unpitched sounds.
As far as I know there’s no clear indication how to handle unpitched sounds in the orchidea-syntax.
A possible solution would be the spectroid centre of the unpitched sounds but this doesn’t satisfy me.
Any advice about the subject would be great. Thank you.
set partials_filter = 0
Any update on this conversion?
Getting contimbre into orchidea would be a dream come true.
I would think this would work. Looking at a couple of the extra datasets available on the orch-idea site, I notice that a couple of them (tam tam and multiphonics) have plenty of samples with no pitch indicated. How accurate the solutions would be I can’t say. But I do know that working from an mfcc database will emphasize timbral match more than the other types of dbs. Not sure how much more, but seems like a good place to start. You might experiment with either of those datasets first to check out the notational results. Renaming the files of a massive dataset like contimbre would be daunting!
Would also be interested in this, let us know if you’ve found a way to realise this 