Ollama can now talk: Gradium's 31M-parameter Phonon brings text-to-speech to Macs
Ollama can now turn text into speech. It arrived Oct 24 in Ollama 0.40.11, and the first speech model is Phonon, a family of small text-to-speech models from Gradium. The default phonon is Phonon 31M English: 31 million parameters, Apache 2.0, so commercial use is allowed. Gradium's 8-bit weights file is 64 MB. You can pipe a chat model's reply into Phonon, and playback starts after the first sentence. There's also an OpenAI-compatible /v1/audio/speech endpoint. The catch: it runs on Apple Silicon Macs only for now, with Windows and Linux "coming soon." The 40M models (English, French, German, Spanish, Portuguese and multilingual) are CC BY-NC 4.0, so noncommercial only.
Try it
ollama run phonon "Hello from Ollama."Needs an Apple Silicon Mac and Ollama 0.40.11 or later. Add > speech.wav to save the audio as a WAV file.