Thanks
Just wanted to say thanks, for the amazing work!
Is using your model locally has any time limit for audio output like 30 sec ?
because i am generating 1 min audio and i have long text but only 30 sec output is generating and for that its compressing my text via increasing speed so much that its not possible to understand.
It was trained to generate audios up to 30s. I'd recommend chunking your input text. We do it in the demo here, if you need reference code!
@bharatcoder , thanks for the support!
this model "model.safetensors " file is nearly 4 Gb . I need a small version of this model so how can i trian this model i just want two languages hindi and english , or may need some sanskrit so can anyone guide me how to retrain this model , a small version of this model , by the way i am new to ai but know some basics of ai , also i know python basics
Hi @abhinandan1111 ,
You can follow the training/finetuning guide from https://github.com/huggingface/parler-tts. You may have to play with the parameters like number of layers, attention heads, and model dim to get it to the size that you want.
Thanks for showing interest in this work.