Supertonic TTS is incredible — sharing our integration results & feedback from real-world usage
Hi everyone!
We’ve been integrating Supertonic TTS into an on-device iOS eBook reader project, and I wanted to share our experience because the results have been genuinely impressive. Among all the offline TTS approaches we’ve tested, Supertonic has provided some of the most natural and stable long-form narration.
Our app is designed to be a fully on-device, privacy-first reader, and Supertonic fit perfectly:
• Runs smoothly via ONNX on modern iPhone/iPad hardware
• Very low latency for paragraph-level inference
• Surprisingly consistent prosody during long-form chapter playback
• Easy integration with EPUB/PDF text pipelines
• Works 100% offline, which is core to our design goals
We tested primarily with classic literature, and—similar to optimization strategies used by other high-end TTS systems—splitting the text into smaller segments and processing them through a queue achieved near real-time performance.
In fact, with this approach, Supertonic reached speeds almost identical to Apple’s built-in system TTS, while delivering noticeably higher audio quality and clarity, especially for longer passages.
I have some feature requests:
Since we plan to keep expanding our TTS features, we’d love to offer some suggestions that could make Supertonic even more powerful for on-device applications:
More voice tones / styles
We’d be excited to see additional optional voice styles.
This would open the door to more dynamic scenarios.Expanded multilingual support
Currently the model works beautifully for English, but we’d love to see future versions support additional languages. Multilingual TTS would make Supertonic an incredible fit for global reading apps.
So far, our testing has mainly been on iPhone 15 Pro and iPad M1 devices, so we don’t yet have detailed performance data for lower-end hardware — insights from others would be especially valuable.
For context, our project is called PageEcho, an on-device AI eBook reader for iOS.
Huge thanks again to the Supertonic team for releasing such an amazing model. We will continue to follow up on future updates.
Thank you for sharing such an awesome app! It's impressive to see Supertonic integrated into a real-world application. We have added six new voices, which you can easily access from this repository. Regarding additional languages, we don't have specific plans yet, but we hope to continue working on this feature, as there have been many requests.
Thanks you for making these amazing voices that can run on commodity devices. I've added support for Supertonic voices to my Read Aloud browser extension, and the new version just got published today. These will probably be used even more than the Piper voices, which came out a year ago now. They're quite good, but Supertonic is next level. My free extension is in use widely in schools, so these voices will be a tremendous boon for students and TTS users. Keep up the great work!