---
language:
- ccp
license: cc-by-4.0
task_categories:
- automatic-speech-recognition
pretty_name: "Chakma ASR Dataset"
tags:
- speech
- asr
- northeast-india
- chakma
---

# Chakma ASR Dataset (ccp)

Speech dataset for Chakma (ccp), a language spoken in Mizoram, India.

## Dataset Description

This dataset contains audio recordings paired with text transcriptions for training and evaluating automatic speech recognition (ASR) systems for Chakma.

### Source

Derived from the ARTPARK-IISc Vaani project (https://vaani.iisc.ac.in/)

## Dataset Statistics

| Split | Samples |
|-------|--------:|
| train | 32,807 |
| validation | 4,513 |
| test | 3,820 |

## Format

- **Audio**: 16kHz mono WAV
- **Text**: Transcriptions in Latin/Roman script

## Dataset Structure

Each sample contains:
- `audio`: Audio recording (16kHz, mono)
- `transcript`: Text transcription

## License

CC-BY-4.0 (Creative Commons Attribution 4.0 International)