How did you manage get the tts Python package into an exe?😅 #49
-
|
I keep running into issues lol |
Beta Was this translation helpful? Give feedback.
Replies: 3 comments 6 replies
-
|
It's just the installer/launcher, not the whole thing - it creates conda environments and installs different packages based on user choice. For an all-in-one package I went with compressing the launcher and the folder with conda envs/dependencies into archives with various "preinstalled" components... It's not very elegant, but easy to use. |
Beta Was this translation helpful? Give feedback.
-
|
Hmm Thx for the info tho 🤔 |
Beta Was this translation helpful? Give feedback.
-
|
On a completely different topic - did you manage to get StyleTTS2 with the replaced phonemiser to work well on Windows? It was unusable for me, mispronounced virtually everything... |
Beta Was this translation helpful? Give feedback.
-
|
Idk if that helps but yeah lol I also had to fix some dependency conflicts with it and the coqui tts pip but was relatively simple lol |
Beta Was this translation helpful? Give feedback.
-
|
I just keep everything in its own conda env. It's bloated, sure, but I'd rather do that than deal with dependency conflicts ; ) The pip package was the one I tried. I thought the problems might be related to the phonemiser (it uses gruut instead of espeak-ng, I think), but if it works ok for you in a Docker container, then it was not the issue. I will probably look into it at some point, but for now, I think XTTS is good enough, especially fine-tuned. |
Beta Was this translation helpful? Give feedback.
-
|
Honestly agreed Xtts is so good that even given the choice on CPU only people still go for Xtts over styletts2 lol |
Beta Was this translation helpful? Give feedback.
-
|
Yes, it's very good, and can be made even better with fine-tuning on 20m of good quality audio, which is simple and not THAT resource intensive. |
Beta Was this translation helpful? Give feedback.
-
|
Honestly yup Pretty insane that all you need is a like a cpu only computer with 16 gb ram lol |
Beta Was this translation helpful? Give feedback.
It's just the installer/launcher, not the whole thing - it creates conda environments and installs different packages based on user choice. For an all-in-one package I went with compressing the launcher and the folder with conda envs/dependencies into archives with various "preinstalled" components... It's not very elegant, but easy to use.