The trend in TTS is increasingly leaning towards a on-line paid solution with AI. It will be difficult to find an open-source solution with a wide range of and/or the possibility of adding voices in different languages and recently written in Java.
Due to the dated open-source software and the large number of changes in the Android OS, once you have found something acceptable in Java code, you will still have to work hard to adapt the TTS software to the current Android security rules for all obsolete software Android components.
Then it is also true that the quality of the open-source voices does not always sound fantastic, which makes a good end result questionable. So its very hard to get a good result.