psibre

P.S. Although now I realize that the statistical cost unitselection variant was actually developed quite a bit earlier (before I joined the project, in fact), around marytts/marytts@260ea4e2b2fa081537513d703a1bbea6afac3de0.

Great to hear you managed to prune away unused transitive dependencies and code! However, please be aware that your efforts at reducing redundant or unused code might be more productively...

Thanks for your thoughts! My two cents: 1. The (possibly) redundant effort I alluded to seems quite real -- many of the problems you mention (duplicate string handling code, etc.)...

No, the confusion is mine! =) When you mentioned `voice-cmu-rms-hsmm-5.2.jar` [above](https://github.com/marytts/marytts/issues/984#issuecomment-633128102), I somehow assumed that you were pulling that in as a dependency for your application, and would therefore be...

You can try the legacy methods detailed [in the wiki](../wiki/New-Language-Support), or the experimental [Gradle plugin](../../gradle-marytts-lexicon-compiler-plugin).

A screenshot is not particularly useful, even when it *does* show the actual error (yours does not). Please attach your `logfile`.

Thanks. Incidentally, it would have been more efficient to zip it and attached it right here. Anyway, I don't see any reference to the error you refer to above. What...

Thanks for the details! > `Error in /home/hiyassat/Trining_EqraFemale1Voice/marytts-5.1/lib/external/bin/HERest` This rather opaque error indicates that something is going wrong with the HTS training scripts (no surprise). But as HTS is not...

Could you please explain what you're asking?

For the lexicon, please see the various lexicon project repositories under https://github.com/marytts. For text normalization, it's trickier, but you can get some leverage out of ICU4J.