msaf.process calling

[bzvew]

When I can msaf.process function, if I have already load the music using librosa.load, can I use that directly?

Also how to use annot_beats and framesync parameter?

[urinieto]

Unfortunately, there's no straightforward way to pass the loaded audio to the algorithms directly. That's something I had in the back of my mind for a while and maybe it's time to add this functionality.

Re annot_beats: You need annotated beats in your dataset to use this. There are annotated beats in, e.g., the Beatles dataset.

Re framesync: You should be able to use this option virtually in any algorithm. This bypasses the beat-synchronous aggregation process, thus making the feature representations (e.g., CQT) much larger, which will make things smaller, yet likely more precise. You may have to tune the parameters of the algorithms accordingly.

Please, let me know if you you need further assistance. Thanks for your comments!

[bzvew]

Great, hope to see that done.