AI-Powered Stem Separation

Splitvoice uses state-of-the-art neural networks including Meta's Demucs and BS-RoFormer to separate mixed audio into individual stems with studio-grade quality.

  • Free: Demucs — 4 stems (vocals, drums, bass, other)
  • Pro: Demucs FT + BS-RoFormer — up to 6 stems (+ guitar, piano)
  • Fine-tuned hybrid transformer models
  • Industry-leading separation quality

Built-in Stem Mixer Pro

Preview each stem individually or create custom mixes by adjusting volume, pan, stereo width, and EQ for each stem in real-time (Pro).

  • Full mixer with Gain, Pan, Stereo Width controls
  • HPF, LPF, and 3-band EQ per stem
  • Export Mixdown with all DSP applied
  • Real-time stem preview

GPU-Accelerated Processing Pro

Leverage your NVIDIA GPU with CUDA acceleration for up to 10x faster processing (Pro only). Complete a full track separation in seconds instead of minutes.

  • NVIDIA CUDA support
  • Up to 10x faster than CPU
  • Automatic GPU detection
  • Falls back to CPU when needed

100% Local & Private

Unlike cloud-based services, Splitvoice runs entirely on your local machine. Your audio files never leave your computer — no uploads, no servers, no privacy concerns.

  • Zero data upload
  • Offline operation
  • No account required
  • GDPR compliant by design

Two AI Models to Choose From Pro

Pick the model that fits your needs. Free includes Demucs for 4-stem separation. Pro adds BS-RoFormer for deeper instrument detail with 6 stems.

  • Free: Demucs — vocals, drums, bass, other
  • Pro: BS-RoFormer — + guitar, piano
  • Switch models on the fly
  • Compare results side by side

Broad Format Support

Work with virtually any audio format. Splitvoice handles everything from high-resolution WAV files to compressed MP3s and modern formats like FLAC and M4A.

  • Import: MP3, WAV, FLAC, M4A, OGG
  • Free export: WAV
  • Pro export: WAV, MP3, FLAC
  • High-resolution audio support (up to 48 kHz / 24-bit)