This repository presents the code used for the submition on the SE&R 2022 Challenge (GPED-CEIA-UFG Team) and the paper Domain Specific Wav2vec 2.0 Fine-tuning For The SE&R 2022 Challenge, based on the Edresson's repository Wav2Vec-Wrapper.
It is important to install the dependencies before launching the application.
Run the following command to install the required dependencies using pip:
sudo pip install -r requeriments
The inference can be made with a language model (using the predict_KenLM.py script) and without a language model (using the test.py script).
The model weights with the best results can be found on the huggingface hub.
- Alef Iury Siqueira Ferreira
If you use this code for your research, please consider citing:
@inproceedings{Alef2022,
title = {Domain Specific Wav2vec 2.0 Fine-tuning for the SE\&R 2022 Challenge},
author = {Alef Iury Siqueira Ferreira and Gustavo dos Reis Oliveira},
pages = {09--14},
url = {http://ceur-ws.org/Vol-3175/paper02.pdf},
year = 2022,
booktitle = {Proceedings of the Workshop on Automatic Speech Recognition for Spontaneous and Prepared Speech \& Speech Emotion Recognition in Portuguese
co-located with 15th edition of the International Conference on the Computational Processing of Portuguese (PROPOR 2022)},
editor = {Ricardo Marcacini and Arnaldo Candido Junior and Edresson Casanova}
eventdate = {2022-03-22}
}
- e-mail: [email protected]