FastPitch Model for Võro
A Võro text-to-speech model developed by the University of Tartu using the FastPitch implementation of the TransformerTTS library. This model has been trained on 2 different Võro speakers. The model is compatible with our TTS worker (v3.*.*) and API workflow. For usage, check these repositories.
This model is also at the time of release the Võro model deployed in Neurokõne and our API.
When using the model, please cite to the following publication:
@inproceedings{ratsep-fishel-2023-neural,
title = "Neural Text-to-Speech Synthesis for {V}{\~o}ro",
author = {R{\"a}tsep, Liisa and
Fishel, Mark},
editor = {Alum{\"a}e, Tanel and
Fishel, Mark},
booktitle = "Proceedings of the 24th {N}ordic Conference on Computational Linguistics ({N}o{D}a{L}i{D}a)",
month = may,
year = "2023",
address = "T{\'o}rshavn, Faroe Islands",
publisher = "University of Tartu Library",
url = "https://aclanthology.org/2023.nodalida-1.73/",
pages = "723--727",
}
The speakers are mapped to speaker ID-s as follows:
sulev: 11
hella: 12
For more information contact ping@tartunlp.ai.
- Downloads last month
- 16