Published February 10, 2024 | Version v4
Dataset Open

The Second DISPLACE 2024 Challenge

  • 1. ROR icon Indian Institute of Science Bangalore
  • 2. ROR icon National Institute of Technology Karnataka
  • 3. ROR icon Indian Institute of Information Technology Dharwad
  • 4. ROR icon Indian Institute of Technology Dharwad

Description

Inspired by the previous session of DISPLACE 2023 challenge, we have  launched the DISPLACE 2024 challenge  (https://displace2024.github.io/). Compared to the first DISPLACE challenge, the current challenge  includes an additional track on automatic speech recognition (ASR) in code-switched multi-accent  conversational scenarios along with speaker and language diarization tracks. We plan to release  both supervised and unsupervised domain matched data for participants to use in model adaptation. To the best of our knowledge, no publicly available dataset matches the diverse characteristics observed in the DISPLACE dataset, including code-mixing/switching, natural overlaps, reverberation, and noise. For this challenge, a natural multi-lingual, multi-speaker conversational dataset will be distributed for development and evaluation purposes. There will be no training data given and the participants will be free to use any resource for training the models. The challenge reflects the theme of Interspeech 2024 - "Speech and Beyond" in its true sense.



The dataset can be obtained by orgainisers by sending the request form and duly signed the terms and condtions

Link for registration for obtaining the data : Registration Link

Files

Track3_ASR_eval_segment_labels.zip

Files (179.5 kB)

Name Size Download all
md5:9a5e088ea4d68bd53485c6e85cbcdd8f
179.5 kB Preview Download

Additional details

Related works

Is continued by
Publication: arXiv:2311.12564 (arXiv)

Dates

Created
2024-01-10
Displace 2024 Challenge Data