HarmonAI Dance Diffusion AEIOU chunking script Howto

DISCORD: / discord
COLAB: colab.research...
Typical settings for chunking .wavs ready to use in fine tuning a 65536 sample size Dance Diffusion model.
AEIOU: github.com/drs...
For more detail on the individual settings please see the discussion on the Discord here
/ discord

[@twobob]

colab.research.google.com/drive/1C4fPoRMimtVJ23gi9TCIrQTYW29_NNmE?usp=sharing Seems it needed a massage

[@user-wn3ro1tc6z]

Hi, thanks very much for making these videos! When I run this colab, I get this error: ''ar' object has no attribute 'nopad'. No chunks are created. I searched the discord and found other people having this same problem, but I didn't find any solutions. Is there anything you might suggest?

[@twobob]

github.com/drscotthawley/aeiou/pull/34

[@twobob]

colab.research.google.com/drive/1C4fPoRMimtVJ23gi9TCIrQTYW29_NNmE?usp=sharing
enjoy

[@fvredits2283]

Hello, thanks for taking the time of making these videos as they are really helpfull for me, a completely begginer, but as the beginner that I am, I have some doubts, my main doubt is, is it okay to have each song separately on wav files like a full album where each song of it is a separated wav, and if so, do I have to ensure myself that there aren't any moments of silence? I think that could affect my final model, correct me if I'm wrong

[@twobob]

Hi. This script removes optionally removes silent chunks. turn STRIP to True for this to happen.

[@fvredits2283]

@@twobob Nice, and thanks for the response :) I'll be trying it out soon