Feature request - Ukrainian model

snakers4 / silero-models

Silero Models: pre-trained speech-to-text, text-to-speech and text-enhancement models made embarrassingly simple

Other

5.01k stars 316 forks source link

Feature request - Ukrainian model #30

Closed egorsmkv closed 4 years ago

egorsmkv commented 4 years ago

🚀 Feature

We would like to have a Ukrainian model for the task of Speech-to-Text.

Motivation

Ukraine has a large population and in the country and there are tons of tasks related to Speech-to-Text.

Additional context

Our group that is based in Telegram ( https://t.me/speech_recognition_uk ) collected a dataset of Ukrainian public speeches/interviews in audio and text formats accessed here: https://mega.nz/folder/T34DQSCL#Q1O8vcrX_8Qnp27Ge56_4A/folder/O3hzlKIJ

We think this dataset will be helpful in the training process.

egorsmkv commented 4 years ago

Also we have own repository where we’re collecting links to datasets: https://github.com/egorsmkv/speech-recognition-uk

snakers4 commented 4 years ago

Hi,

This is exactly the effort I would expect from the community for low-resource languages For now - I will just fit a model on your data as is and share the model via silero-models Then when V3 compact models arrive for all languages, I will consider tuning a Russian model on your corpus

https://github.com/egorsmkv/speech-recognition-uk

Just a few ideas on how to make your repo better:

Add some table with overall statistics
Add some commands (maybe some cli to download your files)
Direct links are always nice

egorsmkv commented 4 years ago

Thanks for your suggestions!

snakers4 commented 4 years ago

Please see https://github.com/snakers4/silero-models/issues/20#issuecomment-723184522

snakers4 commented 4 years ago

I updated the models and purged the CDN cache.

snakers4 commented 4 years ago

It is unlikely that much will change soon, so if everything works let's close the ticket.