---
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