GENERATE SOUNDS Using AI! Generate Sample Packs Locally FREE!

READ THIS FOR INFO! Quick tutorial video on using my python UI to use sample-generator by Harmonai locally! This is mainly for windows but linux SHOULD work aswell.
Atleast 8GB ish of VRAM is recommended for training, or you WILL run into "CUDA Out Of Memory" errors.
If you do not have the required GPU; you can try the notebook in the original repo: github.com/Harmonai-org/sampl...
DISCLAIMER: I am primarily a music producer by craft and program for fun to make tools for myself and my friends; this is not a professionally developed program so there may be bugs.
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
🔗 LINKS:
Python: www.python.org/downloads/rele...
Git: git-scm.com/downloads
GUI Repo: github.com/diontimmer/sample-...
command: git clone github.com/diontimmer/sample-...
Base pretrained model: model-server.zqevans2.workers...
Join the discord! / discord
More info and longer base models are available in this PDF by Markov Pavlov:
drive.google.com/file/d/1nEFE...
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
🎵 Harmonai:
/ harmonai_org
www.harmonai.org/
github.com/Harmonai-org
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
✨ Follow me:
diontimmermusic
dion_timmer
www.soundcloud.com/diontimmer
/ discord
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬

[@plea1938]

10/10 recommend. tried this and my pc made a boom noise 👍

[@surrealstructures1497]

Lol

[@djmarkatmakesmusic]

The best sound ❤️

[@yinyangyt8749]

This all looked really cool. It looks great Dion. I bet you liked it

[@digidope]

Correction: "demo every" and "Checkpoint every" are not epoch values but steps. And there is no reason to leave values on default as the value depends on the size of your dataset. You want to hear demo of a complete epochs and you want to save complete epochs. Number of steps it takes to generate one epoch is dataset / batch size X accumulate batch. If your dataset is not dividable with that number you will create only in-comple epochs. As for other values: For drum hits 65536 is too long, i prefer 32768 and 44.1khz sample rate as most samples are short and 44.1khz. Training and generating samples is then also faster as there a lot less empty in the files. But great tool! I use it all the time.

[@gecktendo]

Very nice tutorial! 😊

[@vocode5604]

Dit is toch niet meer normaal man. 😮

[@AlbiM1]

Damn thats dope

[@made.online2149]

For anyone having issues with this in the future: I highly recommend uninstalling all versions of python, making sure no python environment variables are present, restarting your computer than installing Python 3.10 from the windows store specifically. Restart again, then run the bat file.

[@nylonpython]

Wow, I thought this was going to be a joke.
This is really useful.

[@cartergunter1898]

What minimum gpu do u recommend and do any amd gpus do well

[@highvalue1195]

Dion looks like he’s lost weight! Love you so much ❤️

[@virtualnightgaming]

I panicked when my model was done training and couldn’t turn off the power to my house, so I took out a power line.

[@made.online2149]

I'm having an issue where it generates 1 file and then seems to hang indefinitely despite my gpu still showing as doing work. Is there an updated approach?

[@vincentdiesel8830]

Hello. How to install it? Harmonai dance diffusion?

[@LuciuSOfficial]

You are a legend. Please make more content like this fam 🙏🙏❤❤

[@violantic]

huge

[@user-gh9ex7hc2u]

the Output directory still empty 😯
upd: fixed it.

[@pin9640]

O MA GA DION FACE REVEL

[@Ritorikal]

wizard

[@connermckinney1438]

I love you

[@inplainview1]

Ayo lets go. I've been hoping to dig into AI music tools and perhaps try my hand at some music production.

[@evanfrost7241]

Show some examples