-
https://git.ecker.tech/ aims to provide a place to share my efforts while maintaining true ownership of my code, as I do not trust GitHub.
XMR: 4B9TQdkAkBFYrbj5ztvTx89e5LpucPeTSPzemCihdDi9EBnx7btn8RDNZTBz2zihWsjMnDkzn5As1LU6gLv3KQy8BLsZ8SG
- Joined on
2022-10-10
Per https://github.com/neonbjb/tortoise-tts#v24-2022517:
Removed CVVP model. Found that it does not, in fact, make an appreciable difference in the output.
Long prompts often do not survive the generation to the end of the prompt.
Yeah, that's more of a limitation with TorToiSe itself.
if you use delimiter and random voice, it changes to a…
c27ee3ce95
added update checking for dlas and tortoise-tts, caching voices (for a given model and voice name) so random latents will remain the same
I can probably add it to DLAS's dependencies list at the very least.
I just remember that loading MP3s with tortoise-tts caused a nightmare of esoteric dependency issues that was inconsistent…
I think the crux is that female finetunes are just inherently less likely to turn out good. I've tried Mary/Maria from Silent Hill 2 and it sounds equally shitty. I'll see about finetuning at an…
Your inputs need to be wavs like everything else with the web UI. mp3 support is a muddy mess of dependencies that I have not bothered with at all validating the original repo for.
Right, I forgot they're married to JSON configs. Fixed in mrq/tortoise-tts commit fffea7fc038cc20945aa2208faf91c5434719b69.
Fixed in commit 166d491a987fc33af08997125135ca904943a470.
I didn't get to actually test None for the emotion, just the old behavior with Custom with a blank field.
You shouldn't be running train.bat directly, it's meant to be ran through the web UI.
If you insist on running it directly, you should at least look at the arguments it takes.