Stemdeck

Stemdeck — Free Download. Local Stem separation

StemDeck is a free desktop application for local audio stem separation. Drop in an MP3, WAV, FLAC, OGG/Opus, MP4, or M4A file, or paste a YouTube URL, and StemDeck splits the audio into up to six stems: vocals, drums, bass, guitar, piano, and other. Play them back in a DAW-style multitrack mixer with mute, solo, balance levels, zoom the waveform, loop a region, and export individual stems or a custom mix. Everything runs locally on your own machine without an account, upload, or subscription.

★★★★★
5.0(1 ratings)
Download Stemdeck (Official links)
File size: 272 MB
The latest version of Stemdeck is: 0.15.2
Operating system: Windows, Mac OS, Linux
Languages: English
Developer: stemdeckapp
Price: $0.00 USD (Open Source (Apache-2.0))
  • 6-stem separation via Demucs. The application uses the htdemucs_6s model to separate audio into six distinct stems. It automatically detects the best available Torch device, prioritizing CUDA on NVIDIA graphics cards, MPS on Apple Silicon, and falling back to CPU when no accelerator is present.
  • YouTube and local file import. Users can paste a YouTube URL directly into the import bar or drop a local file onto it. Supported formats include MP3, WAV, FLAC, OGG/Opus, MP4, and M4A. The import process handles both downloaded audio and files already stored on disk.
  • DAW-style waveform editor. The editor renders min/max samples across all stems with shared normalization. Users can zoom in, zoom out, or fit the entire waveform to the screen. A loop region can be dragged on the ruler, and a gold playhead overlay shows the current playback position across stem-aligned lanes.
  • Stem subset extraction. Clicking on stem chips selects which stems to keep for export or playback. When all stems are selected, clicking one chip snaps the selection to only that stem. Subsequent clicks add or remove individual stems from the current subset.
  • Original backing track. When a subset of stems is chosen, a seventh lane appears containing the complement: the full song minus the selected stems. This lane serves as an A/B reference without doubling the audio data, letting users compare their chosen isolation against the remaining material.
  • Downloadable selected mix. The application can export a single mix.wav file containing only the currently selected stems. The stems are summed using ffmpeg amix, producing a ready-to-use audio file that matches the active subset selection.
  • Per-stem mixer. Each stem has a dedicated volume fader, mute button, solo button, and a monitor mode that behaves like solo-only. Mixer state stays synchronized between the preview mixer and the stems sidebar, so changes made in one place appear in the other immediately.
  • Live VU meters per stem. Every stem has a real-time level meter based on post-gain RMS values from Web Audio analysers. The meters include peak hold with a slow falloff, giving a clear visual representation of current and recent signal levels.
  • Song analysis. The program calculates BPM using the librosa beat tracker, estimates musical key and scale with confidence scores from Albrecht-Shanahan profiles, and measures integrated LUFS according to BS.1770 along with sample peak in dBFS.
  • Cancellable jobs. Any running separation or export job can be cancelled mid-pipeline. The runner terminates the active subprocess immediately, deletes the partial job directory, and returns the application to a ready state without leaving incomplete files behind.
  • Library panel. Tracks are organized into folders with drag-and-drop support, a search field, and a trash function for removing items from the library. The panel provides quick access to all imported songs and their separation results.
  • Portable builds. Each Windows and Linux download is self-contained. Users unzip the archive and run the executable without an installer or external toolchain. The in-app updater applies new versions in place for existing installations.

StemDeck began development in 2024 as a project to make high-quality stem separation accessible without cloud services. The application is written primarily in Python for its backend processing and uses web technologies for the interface layer. Development is coordinated through the stemdeckapp organization on GitHub, with contributions from multiple developers focused on local audio processing and cross-platform compatibility. The project has maintained a steady release cadence, with version 0.14.0 addressing a Windows portable startup issue that affected process identity checks.