---
language:
  - bn
  - brx
  - doi
  - gu
  - hi
  - kn
  - kok
  - ks
  - mai
  - ml
  - mni
  - mr
  - ne
  - or
  - pa
  - sa
  - sat
  - sd
  - ta
  - te
  - ur
license: cc-by-4.0
task_categories:
  - automatic-speech-recognition
  - text-to-speech
size_categories:
  - 100K<n<1M
---

# Nirantar

Nirantar speech dataset (22 Indian languages) in Hugging Face format. Source: [AI4Bharat/Nirantar](https://github.com/AI4Bharat/Nirantar).

## Downloading language-wise subsets

Each language is a **separate configuration** (subset), so you can load only one language (like [ai4bharat/Rasa](https://huggingface.co/datasets/ai4bharat/Rasa)):

```python
from datasets import load_dataset, get_dataset_config_names

# Single language (only that language's parquet is downloaded)
ds = load_dataset("adjaysagar/nirantar", "hi", trust_remote_code=True)  # Hindi
ds["train"]  # Dataset for that language

# List available language configs
get_dataset_config_names("adjaysagar/nirantar")  # e.g. ['as', 'bn', 'hi', ...]

# All languages (DatasetDict: config -> Dataset)
ds_all = load_dataset("adjaysagar/nirantar", trust_remote_code=True)
```

## Dataset structure

| Column | Description |
|--------|-------------|
| `audio` | 16 kHz audio (WAV/FLAC) |
| `text` | Normalized transcript |
| `language` | ISO language code |
| `speaker_id` | Speaker ID (if available) |
| `duration` | Duration in seconds (if available) |
| `gender` | Gender (if available) |
| `state` / `district` | Location metadata (if available) |

## Language and metadata statistics

Duration (h) = total audio duration per language. Config name (code) is used in `load_dataset(..., "hi")`.

| Language | Config | Utterances | Duration (h) | GB | Unique speakers | States | Districts |
|----------|-------|------------|--------------|-----|------------------|--------|-----------|
| Bengali | bn | 18,729 | 196.17 | 22.12 | 724 | 1 | 11 |
| brx | brx | 13,220 | 287.36 | 30.85 | 1,061 | 1 | 4 |
| doi | doi | 7,154 | 110.74 | 12.05 | 495 | 1 | 5 |
| Gujarati | gu | 2,174 | 19.27 | 2.07 | 72 | 1 | 4 |
| Hindi | hi | 13,396 | 135.50 | 14.55 | 490 | 4 | 12 |
| Kannada | kn | 8,293 | 94.71 | 10.16 | 530 | 1 | 13 |
| Konkani | kok | 10,604 | 100.15 | 10.75 | 245 | 1 | 4 |
| Kashmiri | ks | 10,247 | 103.57 | 11.17 | 514 | 1 | 10 |
| mai | mai | 21,207 | 241.91 | 25.98 | 726 | 1 | 9 |
| Malayalam | ml | 15,472 | 166.93 | 17.91 | 504 | 1 | 10 |
| Manipuri | mni | 4,133 | 40.85 | 4.38 | 166 | 1 | 3 |
| Marathi | mr | 13,943 | 116.15 | 12.46 | 447 | 1 | 10 |
| Nepali | ne | 26,003 | 249.03 | 26.72 | 780 | 1 | 4 |
| Odia | or | 15,370 | 121.79 | 13.07 | 473 | 1 | 9 |
| Punjabi | pa | 14,762 | 122.92 | 13.19 | 344 | 1 | 6 |
| Sanskrit | sa | 7,332 | 68.33 | 7.33 | 222 | 7 | 17 |
| Santali | sat | 13,503 | 161.29 | 17.30 | 433 | 1 | 8 |
| Sindhi | sd | 4,474 | 26.20 | 2.81 | 240 | 3 | 4 |
| Tamil | ta | 25,747 | 234.50 | 25.20 | 1,242 | 1 | 19 |
| Telugu | te | 13,490 | 217.69 | 23.36 | 767 | 2 | 28 |
| Urdu | ur | 12,219 | 122.05 | 13.10 | 564 | 4 | 10 |
| **Total** | — | **271,472** | **2937.11** | **316.52** | — | — | — |

## Citation

```bibtex
@misc{nirantar2024,
  title={Nirantar: Continual Learning for Indian Languages},
  author={AI4Bharat},
  year={2024},
  url={https://github.com/AI4Bharat/Nirantar}
}
```