Ollama Mac 版支持本地文本转语音模型 Phonon
Ollama for Mac gains support for local text-to-speech models
Ollama 宣布其 Mac 应用现已支持文本转语音生成，首个接入的模型是 Gradium 开发的 Phonon 系列，Windows 和 Linux 支持即将推出。语音可实时生成，用户在模型输出回复的同时就能收听。Ollama 称借助 Apple 的 MLX 框架，在 M5 Max MacBook Pro 上测试时语音生成速度比 Gradium 自身实现快 35%，词错误率与 Hugging Face Transformers 中的参考实现持平。Phonon 支持英语、法语、德语、西班牙语和葡萄牙语五种语言，每种语言有独立模型，另有一个多语言模型，并提供美式、英式英语等不同口音的多种音色。
正文
Ollama, the popular app for running AI models locally, announced that its Mac app now supports text-to-speech generation, with Windows and Linux support coming soon. Here are the details.
Ollama now supports speech models
If you’re not familiar with Ollama, it is a platform that lets users download and run open AI models locally on their Macs, or access more powerful models through the cloud. It is also available for Windows and Linux.
Ollama itself is open-source and has become one of the most popular AI projects on GitHub, with nearly 185,000 stars and around 18,600 forks. You can learn more about it here.
On macOS, Ollama uses Apple’s MLX, an open-source machine learning framework that leverages Apple Silicon’s unified memory architecture, enabling supported AI models to run faster and more efficiently.
Once users get Ollama up and running, they can download compatible open-source AI models from repositories such as Hugging Face and run them directly on their Macs.
This means they can use AI without an internet connection or sending their data to external servers, although performance and the size of the models they can run depend heavily on their Mac’s hardware.
That said, Ollama recently announced support for Phonon, a family of text-to-speech models developed by Gradium, marking the first time the platform supports speech generation.
As Ollama explains, speech is generated in real time, meaning users can listen to AI-generated responses as they’re being produced, rather than waiting for the entire response to finish.
From Ollama:
Ollama can now turn text into speech. The first speech model is Phonon, a family of small, state-of-the-art text-to-speech models from Gradium. Speech models are now available for macOS, with Windows and Linux support coming soon.
The platform says that, thanks to Apple’s MLX framework, it achieved 35% faster speech generation than Gradium’s own implementation in tests on a MacBook Pro with M5 Max, while matching the word error rate of the reference implementation in Hugging Face Transformers.
Phonon supports five languages (English, French, German, Spanish, and Portuguese) with separate models for each language, as well as a multilingual model that supports all five.
Users can also choose from a variety of voices with different accents, including American and British English, depending on the model.
To listen to audio samples or learn more about Ollama’s new speech generation capabilities, including instructions and code examples for using the models through its API, follow this link.
Apple Watch Series 12 | Ultra 4
AirPods Pro 3 | AirPods 5
AirTag (2nd Generation) – 4 Pack
MacBook Neo
Logitech MX Master 4