Where do I download the actual models from?

[zenchantlive]

It seems I am unable to find the models to download to add to the audio_diffusion folder. and, when I run comfy main.py,

When i try to replicate your workflow, this are the options::

I would love to use this!! Thanks for making it. Hope you can help me out.

[NeuralNotW0rk]

The original models from zqevans can be found in the Dance Diffusion notebook

The urls are in this code snippet:

models_map = {

    "glitch-440k": {'downloaded': False,
                         'sha': "48caefdcbb7b15e1a0b3d08587446936302535de74b0e05e0d61beba865ba00a", 
                         'uri_list': ["https://model-server.zqevans2.workers.dev/gwf-440k.ckpt"],
                         'sample_rate': 48000,
                         'sample_size': 65536
                         },
    "jmann-small-190k": {'downloaded': False,
                         'sha': "1e2a23a54e960b80227303d0495247a744fa1296652148da18a4da17c3784e9b", 
                         'uri_list': ["https://model-server.zqevans2.workers.dev/jmann-small-190k.ckpt"],
                         'sample_rate': 48000,
                         'sample_size': 65536
                         },
    "jmann-large-580k": {'downloaded': False,
                         'sha': "6b32b5ff1c666c4719da96a12fd15188fa875d6f79f8dd8e07b4d54676afa096", 
                         'uri_list': ["https://model-server.zqevans2.workers.dev/jmann-large-580k.ckpt"],
                         'sample_rate': 48000,
                         'sample_size': 131072
                         },
    "maestro-150k": {'downloaded': False,
                         'sha': "49d9abcae642e47c2082cec0b2dce95a45dc6e961805b6500204e27122d09485", 
                         'uri_list': ["https://model-server.zqevans2.workers.dev/maestro-uncond-150k.ckpt"],
                         'sample_rate': 16000,
                         'sample_size': 65536
                         },
    "unlocked-250k": {'downloaded': False,
                         'sha': "af337c8416732216eeb52db31dcc0d49a8d48e2b3ecaa524cb854c36b5a3503a", 
                         'uri_list': ["https://model-server.zqevans2.workers.dev/unlocked-uncond-250k.ckpt"],
                         'sample_rate': 16000,
                         'sample_size': 65536
                         },
    "honk-140k": {'downloaded': False,
                         'sha': "a66847844659d287f55b7adbe090224d55aeafdd4c2b3e1e1c6a02992cb6e792", 
                         'uri_list': ["https://model-server.zqevans2.workers.dev/honk-140k.ckpt"],
                         'sample_rate': 16000,
                         'sample_size': 65536
                         },
}

There are also a whole bunch of community-trained models on the Harmonai discord: https://discord.com/channels/1001555636569509948/1025191039352438794

[melMass]

Awesome thanks! IMHO this could be part of the readme (or link to this issue)

[jags111]

This was super helpful and I think adding same to the readme will be very helpful . Any suggestions on the improvement and converting this audio to drive animations will be a wonderful improvement.
Most of the models are available for download from Huggingface page and it is easy to download same.
https://huggingface.co/harmonai