PreTrain Wav2Vec2 in Dhivehi

PreTrain Wav2Vec2 in Dhivehi

There is currently only a multilingually pretrained model for Dhivehi Wav2Vec2. We would like to make a Wav2Vec2 only pre-trained on Dhivehi.

Model

A randomly initialized Wav2Vec2 model

Datasets

  • commonvoice has 18hrs in the last released dataset. [ 32hrs+ if mid 2021 dataset released in time]
  • podcast data [30hr]
  • others []

Available training scripts

FlaxWav2Vec2 will be merged soon: [Flax] Add wav2vec2 by patrickvonplaten · Pull Request #12271 · huggingface/transformers · GitHub and a pretraining script should be relatively easy to be merged.

(Optional) Desired project outcome

The best Dhivehi ASR model

(Optional) Challenges

scraping publicly available Dhivehi audio from various sources

1 Like

Hi, I would like to join this project.

2 Likes

Am interested and would like to join this project

2 Likes

Awesome! Let’s finalize it directly

1 Like

This is a very interesting project. I always wanted to work on speech recognition task. This is a great opportunity to learn and contribute. Looking forward to be a part of this project.

1 Like