auk - Generate and edit speech locally on Apple Silicon
Generates and edits speech locally with AuK using MLX 8-bit models, including voice cloning, TTS, audio effects, separation, and ASR verification.
Tags
Updated: 2026-10-08Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Generate cloned speech
- Synthesize descriptive voices
- Edit spoken content
- Edit sung lyrics
- Adjust pitch and speed
- Adjust volume and emotion
- Modify vocal timbre
- Add or remove nonverbal sounds
- Convert whispering speech
- Remove Mandarin accent
- Enhance and restore audio
- Separate speakers and vocals
- Verify generated audio
- Profile generation stages
Inputs
- Text instructions
- Reference audio
- Source audio
- Voice descriptions
- Edit phrases
- Audio effect parameters
- Speaker cues
- Model weights
- AuK environment paths
Outputs
- Generated WAV audio
- Edited WAV audio
- Verification results
- JSON receipts
- Timing and resource metrics
- Environment diagnostics
Requirements
- Apple Silicon Mac
- Python 3.12 environment
- MLX-compatible GPU
- AuK model weights
- Local Orukeet verifier
- Required audio dependencies