This is the collaborative FAQ for CataRT application and modules for real-time corpus-based concatenative synthesis. CataRT is documented in detail on the IMTR team web site.
You can contribute with questions, answers, and examples, but please notify the community on the concat mailing list.
Contents
- 1 Q: Where can I get information about CataRT?
- 2 Q: How many dimensions can one use to select a grain?
- 3 Q: How do I control the target position for selection by an external controller?
- 4 Q: When is CataRT going to be available on Max 5?
- 5 Q: Do you have plans to make it available on Pd?
- 6 Q: What is the difference between catart.select and catart.select.kdtree?
Q: Where can I get information about CataRT?
A: There are a number of tutorials and help patches within the distribution, and the CataRT page lists a number of documents and articles with in-depth info about CataRT
Q: How many dimensions can one use to select a grain?
A: Any number of descriptors can be used, see the CataRT internal documentation on http://imtr.ircam.fr/index.php/CataRT.
Q: How do I control the target position for selection by an external controller?
A: There are two ways, either via the lcd in 2D using the setpos message with coordinates between 0 and 1 (as in the MIDI-control subpatch), or via catart.select in nD, but you need to map to descriptor values, as explained in the CataRT internal documentation.
Q: When is CataRT going to be available on Max 5?
A: Updated March 2009: CataRT does work on Max5!!! Thanks to the tremendous work of Norbert and Riccardo, the FTM version 2.5 alpha 5 is out, that makes CataRT run fully functional under Max5! There might be a few error messages on startup that can be ignored.
Q: Do you have plans to make it available on Pd?
A: For PD, it would be a lot of work to translate the CataRT application, even though CataRT's basis of FTM&Co. does work on PD. If you're up for it, I'd give you a hand...
Q: What is the difference between catart.select and catart.select.kdtree?
Functionally they appear to be identical. Not even a lot of obvious difference in the code from a quick glance. Can anyone fill me in on why one would use one rather than another?
A: Yes, they're functionally exchangeable. catart.select.kdtree is faster (using mnm.knn instead of mnm.mahalanobis for distance calculation) but doesn't allow for live corpus input (yet), since you have to rebuild the index.