Is unsupervised learning doable with MuBu? Anyone have any experience with it? I’ve taken courses and watched many videos on the subject, but i’m having trouble making the link between theory and practical application (e.g. actual code).
I’m looking to find patterns in segments of audio (or perhaps just midi at first).
Hi
Mubu contains right now principally supervised learning techniques, but we’re planning to add more unspervised learning.
But there are basic methods to segment audio signals for example. What kind of algorithms are you thinking of ?
Fred
How did I never see this response? Well in any case, I’m back at it again.
An example of the sort of unsupervised algorithm I have in mind would be one that can identify repeated patterns in playing. An LSTM model that could provide a percentage of similarity for the current riff compared to riffs that have been played previously, without defining ahead of time what those riffs might be.
I haven’t looked at MuBu as a solution to this project in quite a while. Any advancements since then?