Apppublic

robinhad/ukrainian-stt

sourceHugging Faceupdated 3y agoView on Hugging Face

11likes

App README

🇺🇦🎤 Voice recognition for Ukrainian language

This is a repository with aim to apply various speech recognition models on Ukrainian language.

You can see online demo here: https://huggingface.co/spaces/robinhad/ukrainian-stt. Github link: https://github.com/robinhad/voice-recognition-ua. Source code is in this repository together with auto-deploy pipeline scripts.

🧮 Models

Model name	CER	WER	License	Note
Wav2Vec2	6,01%	27,99%	MIT	Common Voice 8 dataset, `test` set used as validation
DeepSpeech with Wiki LM	12%	30,65%	CC-BY-NC 4.0	Common Voice 6 dataset
DeepSpeech	16%	57%	CC-BY-NC 4.0	Common Voice 6 dataset

Checkout latest releases here: https://github.com/robinhad/voice-recognition-ua/releases/.

If you'd like to check out different models for Ukrainian language, please visit https://github.com/egorsmkv/speech-recognition-uk.

🤖 Training scripts

Guides for training are available in corresponding folders for each model.

Support

If you like my work, please support here: https://send.monobank.ua/jar/48iHq4xAXm

🤝 Attribution

@robinhad - model training. @egorsmkv - organized Ukrainian Speech recognition community. @tarasfrompir - created synthetic 1200h Ukrainian Speech-to-Text dataset. @AlexeyBoiler - hosted Ukrainian Speech-to-Text dataset.