< Back to IRCAM Forum

Creating new datasets / soundbanks for Orchidea

Hi all — new to Orchidea, and wondering if anyone in the community has created databases with their own sampled sounds / if there’s any information on how to do so.

I was considering duplicating the pre-existing TinySOL bank and replacing each .wav with my own samples, but am also wondering if that process could be automated / expedited … Thanks in advance!

Check out the orchidea.db.gen object help file, it shows how to do it.

Basically, you just need to setup a folder with .wav sounds with 16 or 24 bit depth and you use orchidea.db.gen to create the .db file.

To query the database, orchidea expects you to use the convention “instrument-style-note-dynamics-others.wav” to name the files.

Hope this helps!


Wow, thank you @RaphaelVilani — much appreciated! Best of luck with your own music!

@nothingy Were you able to successfully create your own datasets?
Would love to hear about your attempts.

1 Like

@jdeanw I’m sure it could be done, but I decided that the easiest workaround was just to export .xml files from orchidea & “orchestrate” it in my DAW of choice (logic!)

Oh! Wait, does that allow you to use your own samples then?
Sorry if a silly question, brand new to this.

yep @jdeanw — orchidea exports .xml files, which you can open in a DAW as multitrack midi, which can be assigned to samplers or VST instruments of your choice

Im using the windows standalone version, so would I just load my target audio, then export the xml without loading any datasets?
Does the xml translate to multiple instruments?

When you do this, is it possible to preserve microtonal information in the form of pitch bends (assuming each instrument is routed to its own channel)?