Features
Remove vocals and instrumentals from audio/video files
Split stems: drums, bass, electric guitar, acoustic guitar, piano, synthesizer, strings, wind
Lead/back vocal separation with 4 output mixes
Voice cleaning via Lynx neural network (noise, plosives, rumble)
Echo and reverb removal from vocals and recordings
Voice changer with accent and tonality adjustments
Voice cloning from personal audio samples
Lyra offline model: local 7-stem separation in desktop app and VST
Relaxed Queue (free) and Fast Queue (paid) processing
Batch processing on paid plans
Output format selection (MP3, WAV, FLAC, etc.)
API v1 for enterprise and bulk processing
Cross-platform: web, desktop (Win/Mac/Linux), iOS, Android
VST plugin for DAW integration
Upload video files (AVI, MP4, MKV, MOV, M4V)
Text-to-music generation for full tracks and loops
Remixing and style transfer capabilities
Precision audio inpainting and recoloring
Open-weight model release for community experimentation
Curated Model Routing (selects best model for task)
Producer Mode for multi-step workflows
Brand Central for custom brand ID model management
Custom model training on proprietary audio (Enterprise)
Multi-genre support: jazz, drum & bass, lo-fi, techno
High-quality full-length track generation
One-shot sample generator (110GB download, community)
Audio, image, and video generation via Brand Studio
API access available for Enterprise/self-hosted
Legal indemnification with Enterprise plans
Fully licensed training data
