AI Models

We integrate the best open-source speech AI models. All models are available via API and web interface.

Text-to-Speech Models

Chatterbox
Recommended

Zero-shot voice cloning from a short reference sample, in 23 languages. Our default cloning model, the voices in AI dubbing, and our default voice conversion model (Chatterbox VC).

Languages 23
Reference audio 6 to 60 s
VRAM ~5GB
License MIT
Zero-shot Cloning Voice Conversion Multilingual Commercial Use
Kokoro, Piper, VITS and Bark

Four open text to speech engines behind one voice picker. Kokoro 82M (Apache 2.0) is the default; Piper offers only voices whose training data allows commercial use; VITS adds 109 English speakers and Bark 130 presets in 13 languages.

Voices 316
Languages 19
Engines 4
Download MP3, WAV, OGG, FLAC
Multilingual Commercial use allowed
Try text to speech
OpenVoice v2

Tone color converter and cloning model. Still selectable for voice conversion, where Chatterbox VC is now the default because it keeps far more of the target voice.

Voice match in our test OpenVoice v2: 0.17 to 0.35
Voice match in our test Chatterbox VC: 0.64 to 0.75
License MIT

Speaker similarity of converted recordings to a different clip of the target speaker. Two clips of the same real speaker score 0.76 to 0.88.

Voice Conversion Optional
GitHub

Speech-to-Text Models

Whisper Large v3
Most Used

OpenAI's open-source speech recognition model. Transcribes 100 languages, with timestamps and automatic language detection.

Languages 100
Parameters 1.5B
VRAM ~10GB
Language detection Automatic
Default model Multilingual Timestamps
Whisper Large v3 Turbo

A smaller, faster version of Whisper Large v3 for quicker transcripts. Pick it as whisper-turbo.

Languages 100
Parameters 809M
VRAM ~6GB
License MIT
Faster Multilingual

Voice Chat Model

Qwen2.5 7B Instruct

The open-source language model that writes the replies in voice chat, in the persona you pick. Your browser transcribes your speech and a Piper voice reads the reply aloud.

Parameters 7B
License Apache 2.0
Voice Piper
Runs on Our GPUs
Custom Personas
GitHub