ag2003/bhavvani

---
license: apache-2.0
task_categories:
- automatic-speech-recognition
- audio-classification
- audio-to-audio
- text-classification
- text-to-speech
- text-to-audio
language:
- hi
- en
annotations_creators:
- crowdsourced
tags:
- audio
- speech
- speech-emotion-recognition
- automatic-speech-recognition
- audio-processing
- hindi
- multilingual
pretty_name: BhavVani
size_categories:
- 1K<n<10K
configs:
- config_name: default
data_files:
- split: train
path: train.csv
- split: dev
path: val.csv
- split: test
path: test.csv
---

# Exploring Multilingual Unseen Speaker Emotion Recognition: Leveraging Co-Attention Cues in Multitask Learning

This repository contains the BhavVani dataset introduced in the INTERSPEECH 2024 Paper :
[Exploring Multilingual Unseen Speaker Emotion Recognition: Leveraging Co-Attention Cues in Multitask Learning](https://www.isca-archive.org/interspeech_2024/goel24_interspeech.pdf)

Please fill this form for accessing the audio files associated with the BhavVani dataset: [Form Link](https://forms.gle/9AqxS2oY4XVSeH1UA)

## Overview
In our work, we propose the following contributions:
1. We introduce the `CAMuLeNet` architecture for generalizing emotion recognition architectures to unseen speaker distributions using co-attention on features and multi-task learning:

2. We introduce the <b>first-ever</b> Hindi SER dataset - `BhavVani`. The statistics for the same are shared below:

## Citation
If our work was found helpful, please feel free to leave a star and cite our work using:
```bibtex
@inproceedings{goel24_interspeech,
title = {Exploring Multilingual Unseen Speaker Emotion Recognition: Leveraging Co-Attention Cues in Multitask Learning},
author = {Arnav Goel and Medha Hira and Anubha Gupta},
year = {2024},
booktitle = {Interspeech 2024},
pages = {2340--2344},
doi = {10.21437/Interspeech.2024-1820},
issn = {2958-1796},
}
```

## Terms
<b> Commercial and Academic Use: </b>
The database is made available for non-commercial research purposes only. Any commercial use of this data is forbidden.

<b> Redistribution: </b>
The user may not distribute the database or parts of it to any third party.

<b> Publications: </b>
The use of data for illustrative purposes in publications is allowed. Publications include both scientific papers and presentations for scientific and/or educational purposes. In these cases, the identity of the subjects should be protected (i.e., no release of identifiable information of subjects).

<b> Warranty: </b>
The database comes without any warranty. In no event shall the provider be held responsible for any loss or damage caused by the use of this data.