✅ This folder 📁 contains the train, dev and test splits used for all our experiments. Additionally, the folder also contians scripts used to generate those splits. More details can be found here.
The dataset.tar.gz
tar file contains the following csvs: train.tsv,train_random_100h.tsv, train_equi_100h.tsv,dev.tsv,dev_small.tsv,test.tsv and test_small.tsv.
The exact details of how these files are used can be found in our paper.
- Clone the repository
git clone https://github.com/csalt-research/accented-codebooks-asr.git
- Install the required python packages:
pip install -r data/accented-codebooks-asr/requirements.txt
Go to the data folder and execute create_data.sh
script.
cd accented-codebooks-asr/data && ./create_data.sh
The statistics of train, dev and test splits used in our experiments are as follows:
Accent | Train 100h (in hours) | Train (in hours) | Dev (in hours) | Test (in hours) |
---|---|---|---|---|
Australien | 6.95 | 45.36 | 4.33 | 0.46 |
Kanada | 6.79 | 41.13 | 1.16 | 1.21 |
England | 19.51 | 119.9 | 3.22 | 1.65 |
Scotland | 2.69 | 16.21 | 0.23 | 0.16 |
US | 64.12 | 400.1 | 8.32 | 4.87 |
Africa | - | - | - | 1.71 |
Hongkong | - | - | - | 0.52 |
Indien | - | - | - | 0.58 |
Irland | - | - | - | 1.94 |
Malaysia | - | - | - | 0.39 |
Newzealand | - | - | - | 2.11 |
Philippinen | - | - | - | 0.90 |
Singapur | - | - | - | 0.64 |
Wales | - | - | - | 0.27 |
- Darshan Prabhu - M.Tech, CSE, IIT Bombay - Darshan Prabhu
- Preethi Jyothi - Associate Professor, CSE, IIT Bombay - Preethi Jyothi
- Sriram Ganapathy - Associate Professor, EE, IISc Bangalore - Sriram Ganapathy
- Vinit Unni - Ph.D, CSE, IIT Bombay - Vinit Unni
If you use this code for your research, please consider citing our work.
@misc{prabhu2023accented,
title={Accented Speech Recognition With Accent-specific Codebooks},
author={Darshan Prabhu and Preethi Jyothi and Sriram Ganapathy and Vinit Unni},
year={2023},
eprint={2310.15970},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
Distributed under the MIT License. See LICENSE for more information.