File size: 971 Bytes
d56a3df d297f15 d56a3df 4f9df78 6455341 4f9df78 f90b4e5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 |
---
license: cc-by-4.0
datasets:
- KTH/hungarian-single-speaker-tts
language:
- hu
tags:
- text-to-speech
- audio
---
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. The training took about 1 day for the first checkpoint (Step 93000).
Based on the quality of the preview model, we are aiming for 250K Steps.
## Usage
The model 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. |