Ollama’s Mac App Adds Text-to-Speech Capabilities
The update enables real-time, offline audio generation using Phonon models.
Ollama, the popular platform for running artificial intelligence models locally on your own computer, has introduced a significant update. A new text-to-speech (TTS) feature is now available for macOS users, marking a major milestone for the platform. Support for Windows and Linux operating systems is expected to follow in the near future.
What is Ollama and why does it matter?
Ollama is an open-source project that enables users to download and run AI models directly on their own hardware. This approach ensures data privacy, as information is not sent to external servers and does not require a constant internet connection. The project has gained widespread recognition among developers and tech enthusiasts, as evidenced by its popularity on GitHub.
On Apple systems, Ollama utilizes the MLX machine learning framework, which is specifically optimised for the architecture of Apple Silicon chips. This allows models to make the most efficient use of system resources and operate at higher speeds.
Phonon integration and real-time performance
The new functionality is based on Phonon text-to-speech models developed by Gradium. Notably, the system generates audio in real-time, meaning users do not have to wait for the entire text to be processed; they can listen to the output as it is being generated.
During testing on a MacBook Pro equipped with an M5 Max chip, Ollama demonstrated results 35% faster than Gradium’s original implementation. Furthermore, the quality and error rates are fully in line with industry standards.
Languages and availability
Phonon models currently support five languages: English, French, German, Spanish, and Portuguese. Users can select either individual language models or a multilingual version. Additionally, there is an option to choose between various accents, including American and British English variations.
This new feature is available to Mac users with Apple Silicon who are running Ollama version 0.40.11 or newer. Once the application is updated, Phonon models can be downloaded directly from the interface.
This update is particularly relevant for developers and AI enthusiasts looking to integrate AI tools into their own projects. Locally run speech models significantly simplify the creation of personal assistants and automated systems without incurring extra costs or encountering server-side limitations.