Has anyone successfully set up tortoise-tts before?
It's for open source ai generated speech. https://github.com/neonbjb/tortoise-tts
That thing is like an 11 out of 10 install difficulty. I hate github projects that are really difficult to get working.
I'm using Debian. So many fucking issues. It probably can't work on Debian but I'd really like to know what system people are using who have been successful in getting this to actually work.
They have a container and instructions on how to build and run it. Where are you getting stuck?
Could you be more specific on what issues you are having? What Debian version are you on?
I did some time back. I don't recall it being incredibly difficult to get running. What hardware are you trying to run it on?
I had it running on Windows (no container) a while back. Wasn't particularly difficult at that time, at least.
Can't give any advice here though, since all we've been given to work with is an OS.
Don't use that repo, it's old/broken. Use one of these two:
The docker installation for the first one is broken just FYI but I guess I should be trying to use whichever one is the most up to date for higher chances of success.
I struggled to get it set up last night, eventually I stumbled across issue 796 on the github which had the solution. https://github.com/neonbjb/tortoise-tts/issues/796#issuecomment-2204846504
Seems there's an issue with the dependencies currently and all versions of installing tortoise-tts from the existing instructions is doomed to failure without manual intervention.
As mikejgrecojr commented in the linked github issue, the fix for running via docker is:
Try updating your Dockerfile by adding in scipy to the conda install and specifying version 1.13.1 on line 31 like below. That worked for me:
&& conda install pytorch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 pytorch-cuda=12.1 scipy=1.13.1 -c pytorch -c
I have it working on Debian, it wasn't THAT hard, but I never got it to work with the GPU so it was SUPER slow. I've since found XTTS2 which set up super easy, comes with a web GUI and just supports my GPU out of the box.
Coqui xtts is broken right now, at least on Debian. I did not try xtts2. Yet turns out I already tried that one. After my computer is done being frozen from my latest hacking attempt to make a borked tts python maybe work, I'll add whatever the error is.
What's the usecase? I fount tortoise-tts very niche tbh