Skip to content

Dataset · HumynLabs

Noisy airline support audio

HumynLabs/airline-customersupport-Hinglish-audio

Deliberately noisy audio recordings with a Hinglish phrase to test audio model robustness.

What's inside

The dataset contains intentionally low-grade ("B-grade") audio recordings: noisy, imperfect, or otherwise suboptimal samples. Participants utter an airline customer support phrase in Hinglish regarding flight delays and booking the next one. Size is under 1K records, modality is audio; metadata indicates the en language, but the phrase text is in Hinglish. The data is designed to test model robustness to degraded input.

What you can do with it

  • Testing ASR on noisy speech
  • Testing audio classification under degradation
  • Evaluating airline support voice agent robustness
Tasks
Audio classification
Languages
English
Size
up to 1K rows
Downloads this month
190
Likes
438

How to load it

from datasets import load_dataset

ds = load_dataset("HumynLabs/airline-customersupport-Hinglish-audio")