egorsmkv / ukrainian-accentor

Add accents to words in the Ukrainian language

Accentor for Ukrainian

Overview

This repository contains a PyTorch model to make accents in Ukrainian words.

The model was trained using this notebook in notebooks/ folder

Alternatives

• Ukrainian word stress by Oleksiy Syvokon

Installation

pip install git+https://github.com/egorsmkv/ukrainian-accentor.git

Demo

python inference.py

With stress: Я́ співа́ю ве́селу пі́сню в Украї́ні
With pluses: +Я спів+аю в+еселу п+існю в Укра+їні

Or you can use the library like the following:

import ukrainian_accentor as accentor

# Using GPU
# accentor.cuda()
# Back to CPU
# accentor.cpu()

text = "Я співаю веселу пісню в Україні"

stressed_words = accentor.process(text, mode='stress')
plused_words = accentor.process(text, mode='plus')

print('With stress:', stressed_words)
print('With pluses:', plused_words)

Development

If you want do train your own model clone this repo and run:

pip install -r requirements.dev.txt

Attribution

Training script is a derivative of word_stress_pytorch.ipynb by dsakovych

Sentence support implemented by code from tokenize-uk by lang-uk