Ollama now reads aloud: Gradium's Phonon runs locally on Mac, English is under Apache 2.0, and the other four languages are for noncommercial work only

Phonon in English is under Apache 2.0. In French, German, Spanish and Portuguese it is under CC BY-NC 4.0, and Ollama says it plainly: noncommercial use only. There is no Bulgarian.

Ollama now turns text into speech with one command: ollama run phonon. Phonon is a family of small models that Ollama calls state-of-the-art. It talks but does not listen: the post has no speech recognition.

For a builder the interesting part is the door to the code. Phonon answers on the OpenAI-compatible endpoint, so code that already uses the OpenAI SDK only needs a new base URL, Ollama writes. You can pipe a chat model's reply straight into Phonon: it is spoken as the text arrives, and playback starts after the first sentence.

One family, two licenses: English is free, the other four languages are not for commercial work.

Here is how it splits. The multilingual variant is one of the 40M models, so it carries the same license as the four languages and is not for commercial use either.

If you are on Windows or Linux, you wait: Ollama only says it is coming soon, with no date. If you put a voice into something you sell, for now you get one language: English.

The visual is generated code-art, no third-party images.