---
license: cc-by-4.0
datasets:
- KTH/hungarian-single-speaker-tts
language:
- hu
tags:
- text-to-speech
- audio
inference: false
---

This vits model was trained on the [KTH/hungarian-single-speaker-tts](https://huggingface.co/datasets/KTH/hungarian-single-speaker-tts) dataset.

## CSS10 Hungarian: Single Speaker Speech Dataset
The corpus consists of a single speaker, with 4515 segments extracted from [this](https://librivox.org/egri-csillagok-by-geza-gardonyi/) single LibriVox audiobook. It consists about 10 hours of audio data.

## Training
The model was trained on a single RTX 3090 GPU for 3 days, 200K steps with a batchsize of 16. 
We saved some checkpoints with the optimizers, so the model could be train further, however we didn't find any noticable effect after step 150K.

## Usage
The model `diana_final.pth` can be used with [JayWalnut's git repo](https://github.com/jaywalnut310/vits), but you have to modify the `text/cleaners.py` file to contain our `hungarian_cleaners` method.
We provided the necessary files in our repo to do so.