Free AI stem splitter that works with videos

Aug 19 in AI 👾

Hey team, recently I discovered UVR5. A free tool that uses AI models to remove vocals from songs and split stems (6 different tracks). Surprisingly it also works with videos!

I've been mainly using this tool for two things:

Practising a specific instrument: I want to improve my drum programming skills. For that, I split the stems, imported them into my DAW, and muted the drum track. Then I write the drums myself. During this process, I often do the reverse process. I mute all the tracks but the drums, to see how I compare to the existing drums.
Composing for trailers: After splitting the stems of a video trailer and importing them on my DAW, I mark the breaks (intro, main, climax, etc...) and then I start composing, strictly following those marks. After I'm done (that never happened, I'm still to finish a track using this method 😁) I see how my track fits in that existing trailer.

If you are interested in doing the same as me, download UVR5. After that, go to the settings and download the htdemucs_6s (Demucs) model (This AI model splits into 6 stems).

After the model is downloaded, choose the DEMUCS process model, and the model you just downloaded.

Click the select input, and select the audio/video track you want to split. And wait... this process might take some time depending on your machine.

Let me know if I can help you with something!

Cheers.

8 comments