README.md

metadata

language:
  - hi
license: cc-by-4.0
task_categories:
  - automatic-speech-recognition
task_ids:
  - speech-recognition
pretty_name: Common Voice Hindi – Mini ASR
source_datasets:
  - mozilla-foundation/common_voice
size_categories:
  - 1K<n<10K

Common Voice Hindi – Mini ASR Dataset

This dataset is a 1000-sample subset of the Mozilla Common Voice Hindi corpus, prepared for Automatic Speech Recognition (ASR) experiments.

Dataset Structure

hf_mini/ ├── train/ │ ├── validated.tsv │ └── clips/ └── test/ ├── test.tsv └── clips/

Columns

• path — relative path to the audio file
• sentence — ground truth transcription

Audio Details

• Format: MP3
• Sampling rate: 16 kHz (decoded at runtime)

License & Attribution

This dataset is distributed under the CC-BY-4.0 license.

Original data source: Mozilla Common Voice (Hindi)
© Mozilla Foundation