README.md

metadata

license: cc-by-nc-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

Please fill this form for accessing the audio files associated with the BhavVani dataset: Form Link

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 first-ever 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:

@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

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

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

Publications: 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).

Warranty: 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.