stemdeckapp/stemdeck — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a free local desktop app for musicians called StemDeck. I want someone to drop in an MP3 or WAV, or paste a YouTube link they have the right to use, and have the app split the song on their own computer into vocals, drums, bass, guitar, piano, and other. No accounts, no uploads, no subscription, and make it clear that the audio stays private.

After processing, show a clean waveform workspace with separate lanes for each stem. Let the user play, pause, zoom, loop a section, mute, solo, change volume, and see simple level meters. They should be able to pick only certain stems, hear the original backing track against them, download each stem, or export a custom mix.

Also add a small library area for saved tracks with search, folders, drag and drop, and trash. Show useful song info like BPM, key, loudness, and peak level. Use Demucs or whatever current local model is best, and look up current docs online if needed.

Want more depth? Deep Reverse