search

erew123 / alltalk_tts

AllTalk is based on the Coqui TTS engine, similar to the Coqui_tts extension for Text generation webUI, however supports a variety of advanced features, such as a settings page, low VRAM support, DeepSpeed, narrator, model finetuning, custom models, wav file maintenance. It can also be used with 3rd Party software via JSON calls.
GNU Affero General Public License v3.0
1.09k stars 115 forks source link

V2 Beta fresh install will continue to download models as if the first time load. #329

Closed beeblebrox closed 2 months ago

beeblebrox commented 2 months ago

Describe the bug After a fresh install and first time startup if you select xtts it will download the models. Then if you restart (ctrl-c, rerun start_alltalk.bat) after the server is up and running and even after changing a global setting (in my case enabling rvc) the next run will still have a first time prompt for the model to download and then download the models a second time for xtts. After I downloaded the piper models it stopped prompting me on other restarts.

To Reproduce Steps to reproduce the behaviour: See above description.

Desktop (please complete the following information): AllTalk was updated: 15th August 2024 at 08:27 Custom Python environment: ish, had to change requirements_parler.txt to have git+https://github.com/huggingface/parler-tts.git@862f84181589ea980d41d279ec53841bc8e178d2^ to workaround conflicting dependency issues (different issue)

Text-generation-webUI was updated: NA

Additional context This is from a fresh install, and seems very reproducible if selecting xtts on that first prompt and don't download any other models before the next restart.

erew123 commented 2 months ago

Hi @beeblebrox

Not sure why it wasnt flagging that first run had completed, though I have doubled the code that marks the firstrun has been completed https://github.com/erew123/alltalk_tts/commit/57a6b3ba3936ecde7362303104c36de3f77afd28

image

So that should definitely catch it.

Re the Parler/Transformers messages, I have covered that in this ticket here https://github.com/erew123/alltalk_tts/issues/332 so wont repeat here.

Thanks