Ollama's Mac App Now Supports Text-to-Speech

Ollama, a popular open-source platform for running AI models locally, has announced that its Mac app now supports text-to-speech. Windows and Linux versions are coming soon. The project has nearly 185,000 stars on GitHub.

The first supported voice model is the Phonon series from Gradium. It is small and runs offline on Apple Silicon Macs without any cloud service, which works well for privacy. On macOS, Ollama uses Apple's open-source MLX framework to boost performance.

Official tests on a MacBook Pro with an M5 Max chip show that Phonon generates speech 35% faster than Gradium's own implementation, while keeping a similar word error rate. It can even speak while the answer is still being generated.

Phonon currently supports English, French, German, Spanish, and Portuguese. Users can pick accents like American or British English, making local AI conversations feel more natural.