---
dataset_info:
  features:
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
  - name: text
    dtype: string
  - name: english
    dtype: string
  splits:
  - name: train
    num_bytes: 3587311906.809
    num_examples: 26453
  download_size: 3332905202
  dataset_size: 3587311906.809
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
task_categories:
- automatic-speech-recognition
- translation
language:
- ig
- en
---


`igbo-dict` is an Igbo text-audio dataset that includes the following:
* 25,500 single word audio recordings for each dialectal word variation
* 25,000 single Igbo sentence audio recording for each Igbo-English sentence pairing

**The original audios were converted to 16kHz WAV.**

Referenced in [The IgboAPI Dataset: Empowering Igbo Language Technologies through Multi-dialectal Enrichment
](https://arxiv.org/abs/2405.00997)

## Citation
If you want to cite this dataset you can use this:

```bibtex
@misc{igbo-dict-16khz,
  title={Igbo dataset},
  author={Jimenez, David},
  howpublished={\url{https://huggingface.co/datasets/deepdml/igbo-dict-16khz}},
  year={2025}
}
```