OmniVoice

OmniVoice is an Apache-2.0 voice model offering text-to-speech, zero-shot voice cloning from a few seconds of audio, and voice design from a text description. It runs locally on NVIDIA GPUs, Apple Silicon, or CPU (with an optional cloud tier), covering 646 languages from a single unified model at roughly 45x real-time.

Best for: Developers who want open, local voice cloning and TTS with broad language coverage

• ✓ Text-to-speech, voice cloning, and voice design
• ✓ Runs locally on NVIDIA GPU, Apple Silicon, or CPU
• ✓ 646 languages from one unified model
• ✓ Zero-shot cloning from 3–25s of audio
• ✓ Apache 2.0 — free for commercial use
• ✓ ~45x real-time synthesis