Support useTextToSpeech with Supertonic, a lightning-fast, on-device multilingual text-to-speech

[StevePhuc]

Problem description

https://github.com/supertone-inc/supertonic

Some quick info about the modal

⚡ Blazingly Fast — Low-latency, real-time synthesis across desktop, browser, mobile, and edge — fast enough to turn an entire webpage into audio in under a second
🌍 31-Language Multilingual
🪶 99M-Parameter Open-Weight Model
📱 Edge-Device Ready — Runs locally on desktop, mobile, browsers, and resource-constrained hardware like Raspberry Pi or e-readers, with zero network dependency, complete privacy, and no GPU required
🛠️ Multi-Runtime SDKs — Ready-to-use examples through ONNX Runtime across Python, Node.js, Browser (WebGPU), Java, C++, C#, Go, Swift, iOS, Rust, and Flutter

Proposed solution

add to support modal in https://docs.swmansion.com/react-native-executorch/docs/hooks/natural-language-processing/useTextToSpeech#supported-models

Alternative solutions

No response

Benefits to React Native ExecuTorch

support more language than current tts solution

Additional context

No response

[msluszniak]

Hi @StevePhuc, thank you for submitting an issue. Supertonic looks like a great addition which will cover more than 20 new languages in the library. We will check it out, try to export to ExecuTorch format and reach out to you with feedback if we can add this model. 😄

[msluszniak]

I see one pretty big blocker here. There is no PyTorch model of supertonic 3, only ONNX format. Huggingface model card of the Supertonic 3 states that explicitly:

This model was trained using PyTorch, which is licensed under the BSD 3-Clause License but is not redistributed with this project. See the PyTorch license for details.

So as long as we don't have any format suitable for export to *.pte format, we won't be able to export it and support in our library :/

Marking it blocked. If you'll see something like this: https://github.com/chantysothy/supertonic-pytorch but for Supertonic 3, please reach out to us.