fine-tuning
how to fine tuning this model. i tred with https://github.com/ylacombe/finetune-hf-vits this Introduction, but not working. Problem:OSError: facebook/mms-tts-eng does not appear to have a file named preprocessor_config.json...
How exactly do you set it up? This model is a bit strange. It takes effort to get it to generate sound from text. With the help of a chatbot, I got it working with a lot of effort.
Although with a huge delay. To remove the checkpoint from this script you have to delete something that blocks it in this file "convert_original_discriminator_checkpoint.py". Unfortunately I don't remember what exactly. In my space there is the training code, but it is for my native language and there is the checkpoint only for it converted into pth format. My code is much shorter, but there is currently no script to clean up the model from the excess after training. I will post the code for testing soon. I have it for cleaning, but it is old and I don't remember what it lacks to work. Of course, it will have to be adjusted according to the parameters of the model that will be trained. I personally don't use "uroman", telemetry and need to log in to this site. Here is a link - "https://huggingface.co/batvanio12/Bg-TTS-Vits-MMS-checkpoint". Hope it helps. 😁
Here is the new improved version
https://huggingface.co/batvanio12/Bg-TTS-Vits-MMS-checkpoint/tree/main/Fine-Tuning