Credits

Character voices are AI-generated

Every voice that reads your scene partners' lines is synthetic speech made by a text-to-speech model. None of them is a recording of a real actor. The voices were designed from synthetic reference voices, not by cloning real people.

Standard voices

Standard voices run on your device using Pocket TTS by Kyutai, licensed under CC BY 4.0. We run the ONNX conversion by KevinAHM unmodified. The voices themselves are voice-reference data we derived from Google Gemini's synthetic voices.

Premium voices

Premium voices are generated by Google Gemini text-to-speech and xAI Grok text-to-speech.

Open-source software

• pocket-voice by Hemanth.HM (MIT): the in-browser Pocket TTS runtime, which we modified.
• ONNX Runtime Web by Microsoft (MIT).
• sentencepiece-js (Apache-2.0), including Google's SentencePiece (Apache-2.0).