Today, I downloaded the Piper, Neural TTS app on my iPhone. A couple of observations:
1. Due to what I believe is it's use of the length scale parameter to change speaking rate (I don't think there is actually another way to do this in Piper itself, pretty sure the Sonata addon for NVDA uses Sonicc), the speaking rate slider in Voiceover is actually the inverse of what you would expect. The higher the speaking rate, the slower the speech.
2. I tried to import a local voice that isn't in the community repository, and probably never will be. There is an 'import from files' button, which theoretically allows you to import a .onnx and it's accompanying .json file for a voice, but when I point to a folder and select files, then tap open, nothing at all happens, as far as I can tell. I am brought back to the main screen of the Piper app. I tried importing from a local folder on my phone, as well as Dropbox.
Has anyone experienced this behavior?