152334H / tortoise-tts-fast

Fast TorToiSe inference (5x or your money back!)
GNU Affero General Public License v3.0
771 stars 179 forks source link

Download models once #18

Closed demianvle closed 1 year ago

demianvle commented 1 year ago

First, thank you for working on this. I believe its awesome you are trying to keep this open and available locally for users who don´t want to depend on cloud, third party services.

My request/questions: can you follow a similar path to Automatic1111 SD´s where models are downloaded and added to a folder once, manually, vs the models being downloaded every time TTS runs? I also like the webui idea, I just would prefer if running tts-fast could be done without internet connection.

Thanks!

demianvle commented 1 year ago

I´m also getting this: AppData\Local\Programs\Python\Python310\lib\site-packages\torch\amp\autocast_mode.py:202: UserWarning: User provided device_type of 'cuda', but CUDA is not available. Disabling warnings.warn('User provided device_type of \'cuda\', but CUDA is not available. Disabling')

I have a GTX1060, 4gb, I was still able to generate some audio but it ran on the CPU and took forever. could this be on my end?

152334H commented 1 year ago

My request/questions: can you follow a similar path to Automatic1111 SD´s where models are downloaded and added to a folder once, manually, vs the models being downloaded every time TTS runs?

uhhhhhhhhhhhh this is supposed to happen by default. Are you using do_tts.py or something else?

UserWarning: User provided device_type of 'cuda', but CUDA is not available.

Either CUDA is not installed or you installed the pytorch package without CUDA. If it's the latter, check the pytorch website on instructions to install pytorch with CUDA.

demianvle commented 1 year ago

My request/questions: can you follow a similar path to Automatic1111 SD´s where models are downloaded and added to a folder once, manually, vs the models being downloaded every time TTS runs?

uhhhhhhhhhhhh this is supposed to happen by default. Are you using do_tts.py or something else? (This I guess I didn´t check fine, seems to be ok)

UserWarning: User provided device_type of 'cuda', but CUDA is not available.

Either CUDA is not installed or you installed the pytorch package without CUDA. If it's the latter, check the pytorch website on instructions to install pytorch with CUDA.

I reinstalled Cuda, uninstalled pytorch and reinstalled and its working now. Sorry to bother :p Thank you and keep up the great work. I think keeping AI tools open and locally run is key for our future. I´m only sorry I can´t contribute/help much. I will download test and spread the word for sure.