Credits

RockChat:AI stands on the work of the people and projects below. Thank you.

The rock voices

The voices are Pocket TTS, a text-to-speech model by Kyutai (kyutai.org). Model weights: Creative Commons Attribution 4.0 (CC BY 4.0).

• marius, javert: the voice-donations collection, CC0 (public domain).
• alba: recordings of Alba Mackenna, CC BY 4.0.
• azelma, eponine, fantine: the CSTR VCTK Corpus, University of Edinburgh (Junichi Yamagishi, Christophe Veaux, Kirsten MacDonald), CC BY 4.0.

Speech runs through pocket-tts-ios, an iOS port of Pocket TTS in Rust (MIT license), by UnaMentis and babybirdprd, packaged for this app by ahalekelly. It builds on Candle by Hugging Face.

The language models

The rock thinks with an open language model that the app downloads once from Hugging Face. The built-in choice is Google's Gemma 4, converted for Apple's MLX by community contributors. Gemma 4 is released by Google under the Apache License 2.0.

Everything else

The full list of libraries and their licenses is in the app: open settings, tap about at the bottom right, then credits and licenses.