Neural audio timbre transfer with RAVE trained on Amen breaks using nn~ in Pure Data

[@marvbordello6047]

dope

[@ShihWeiChieh]

hey, i really wanna install nn~ in max and learn about how to train my model with giga byte data. do you use google collab? or is there a way to not to? thank you so much and super envy!

[@martsm_n]

Colab or Kaggle are solutions if you can't or don't want to train on your local machine. I've set up notebooks for these platforms, you can find them here: github.com/devstermarts/Notebooks

[@ShihWeiChieh]

@@martsm_n thanks a lot! super neat notebook and i think the training is successfully initiated! now 22 epoch. Do you know how to know the max number of epoch?

[@martsm_n]

@@ShihWeiChiehtraining time is highly dependent on dataset, parameters and purpose. I can recommend joining the RAVE discord channel and ask around there for hints and intuitions.

[@ShihWeiChieh]

@@martsm_nok I will, thank you and keep up the good work!

[@ShihWeiChieh]

i got some helps from the discord and i have my first .ts file trained. I guess I have to train my prior model next for max/msp and pd usage (because the first .ts file is not compatible with my nn~ object). So i started the MSPRIOR notebook of yours then i stuck with the training cell: "ValueError: cannot reshape array of size 131072 into shape ()". Can you help me with this?

[@MFRSIAM]

how are u doing latent space manipulation ? i see u are using a resnet but am not sure whats going on here ?

[@martsm_n]

This is actually footage from very early and cautious interventions. As far as I can recall, it's not much more than altering the spread of latents per dimension given - in PD that'd be amplitude modulation of a signal value. You can find more videos and techniques of latent intervention in my Neural Audio playlist: kzbin.info/aero/PLmNfQif1XJQ0aNdrqk8HxvwUxq90xc9al

[@davidevitturini5837]

Is the model available anywhere?

[@martsm_n]

No. I didn't release it. You can train it quite easily yourself, however.

[@davidevitturini5837]

@@martsm_n did you used V1 or V2? On V2 and around 6hrs dataset with 2 3060 12gb it's taking around 1 week for 1 M epochs (I'm using a batch size of 36) [do you have a discord for discussing?]

[@martsm_n]

@@davidevitturini5837 This one is a V2 model with default regularization. Another V1 model has been used in this video: kzbin.info/www/bejne/gpixnoyLf8aGn5Y

[@davidevitturini5837]

@@martsm_n thanks, really enjoyed your creativity