Spaces:
Running
on
Zero
Orpheus Music Transformer
Hey guys!
I hope ya all doing well :)
Just pinging you to bring your attention to the finished version of my latest creation: Orpheus Music Transformer
IMHO the model turned out very well and it plays very well too.
I would really appreciate any feedback and likes. This will help a lot.
Also, if any of you can help promote the model in any way, I will appreciate it as well very much :)
Here are the links for your convenience:
Orpheus Music Transformer main demo space
Orpheus Music Transformer Collection with all demo spaces of all features
Thank you very much in advance for your time and assistance! :)
Sincerely,
Alex
this rules, just catching up now
Hey, just wanted to say that I really enjoy your models and datasets! Since I like both music and coding, this is pretty much the perfect match. I'm currently planning on making a musical transcription model, and maybe you could use it to train a new model on a lot more songs. Right now it's in the planning phase, and I have exams coming up, but I'll try to work on it. Either way, I'm going to announce you when I finish it. (expect 8 months from now or more 😢). What do you think?
@vericudebuget Hey!
Thank you for your kind words :) It means a lot to me :)
Yes, definitely let me know. I would be happy to try it out and use it in my projects :)
What kind of transcription model do you plan to make? Solo Piano or multi-instrumental?
Alex.
I plan to make a multi instrumental one, that uses a custom format, (I'm thinking of calling it RSSR or RSR, refined sound symbolic representation), and it will be similar to MIDI, but with much more robust modifiers, eg: for vocals, in will include every micro pitch, tremolo, attack, breathiness, etc. (you can look up melodyne for a rough ideea) And different stuff for different instruments, like violin, guitar, Pretty much everything. + effects, (reverb, echo, delay, flanger, chorus, etc), so that you could get the best out of it. But don't worry, you will very easily be able to convert these files to MIDI, mostly use this for instrumentals. Also, right now I'm making a DAW, from scratch (It will be free!), and obviously it will be compatible with this sound format (this is where I got the idea). So far, the DAW has a bunch of features, + some pretty decent instruments (A bunch of pianos, and most of them have 24 bit WAV samples, 16 velocity layers, stereo recordings with studio mics, and a bunch of other features) a single piano instrument, (e.g., the Yamaha C4) is over 1.4 GB, so it's pretty beefy. But keep in mind, it is nowhere near finished. Expect at least 2y before the first version comes out. The upcoming exams and this DAW are what are keeping me from developing the transcription model, but if God is willing, I should be able to start working on it by next year. Is there anything you would recommend me to do? What do you think about this?
@vericudebuget Definitely sounds interesting and promising. I could definitely use good multi-instrumental transcription model in my projects.
I highly recommend checking out Google MT3 model and a similar spin-off YourMT3 model:
https://huggingface.co/spaces/mimbres/YourMT3
Get in touch with the guy who made YourMT3. He is friendly and he can advise you how to make a good transcription model.