StemDeck, a free, open-source and local AI stem separator

StemDeck is a free and open-source local AI tool designed to separate audio into individual stems, such as vocals, drums, bass, guitar, piano, and other elements. Users can upload MP3, WAV, FLAC, OGG/Opus, or MP4 files, or paste YouTube URLs directly into the import bar, and the tool will process the audio locally on their own machine. StemDeck does not store, cache, or redistribute any downloaded content, ensuring that everything happens locally and no data leaves the user's device.

Unlike cloud-based stem-separation services like Moises and LALAL.AI, StemDeck offers a free alternative with no account creation, no quotas, no uploads, and no subscription fees. It provides a DAW-style multitrack mixer where users can mute, solo, balance levels, zoom waveforms, loop regions, and export individual stems or custom mixes.

The tool also features a waveform editor with sample rendering, zoom in/out, fit, loop drag, and a gold playhead overlay. Users can extract stem subsets, choose which stems to keep, and view a complement lane containing the full song minus the selected stems for A/B reference.

StemDeck offers a per-stem mixer with volume faders, mute, solo, and monitor controls, along with live VU meters, per-stem analysis (BPM, key, scale, confidence, LUFS, and sample peak), and a state synchronization feature between the preview mixer and the stems sidebar. The tool is built on Python 3.12, utilizes the Demucs 6-stem neural network for separation, and supports YouTube audio fetching via yt-dlp, transcoding and mixing with FFmpeg, BPM detection and key analysis through librosa, and loudness measurement using pyloudnorm (ITU-R BS.1770).

StemDeck is compatible with macOS and Windows desktop shells, with a self-contained .app and DMG package that includes its own runtime, FFmpeg, and the Demucs model.