changelinglab/tusom2021-pr

---
license: mit
configs:
- config_name: default
data_files:
- split: test
path: data/test-*
dataset_info:
features:
- name: id
dtype: string
- name: audio
dtype: audio
- name: transcript
dtype: string
splits:
- name: test
num_bytes: 134023918
num_examples: 2255
download_size: 130816848
dataset_size: 134023918
task_categories:
- automatic-speech-recognition
tags:
- phone-recognition
- low-resource-language
pretty_name: Tusom2021
---

# Tusom2021

> [!NOTE]
> If you use this dataset, please remember to cite:
> ```
> @inproceedings{mortensen2021tusom2021,
title={Tusom2021: A Phonetically Transcribed Speech Dataset from an Endangered Language for Universal Phone Recognition Experiments},
author={Mortensen, David R and Picone, Jordan and Li, Xinjian and Siminyu, Kathleen},
booktitle={Proc. Interspeech 2021},
pages={3660--3664},
year={2021}
}
> ```

[Tusom2021](https://github.com/dmort27/tusom2021) is a speech dataset from East Tusom, an endangered Tibeto-Burman language of Northeast India.
The original dataset provides glosses and phonetic transcriptions (with versions that including tones), while we include only the transcriptions without tone marks.
Other than that, the only change is that we host it on Hugging Face for ease of use 🤗

This dataset is included as `pr-tsm` in our phone recognition benchmarking toolkit, [💎PRiSM](https://github.com/changelinglab/prism).

```
@misc{prism2026,
title={PRiSM: Benchmarking Phone Realization in Speech Models},
author={Shikhar Bharadwaj and Chin-Jou Li and Yoonjae Kim and Kwanghee Choi and Eunjung Yeo and Ryan Soh-Eun Shim and Hanyu Zhou and Brendon Boldt and Karen Rosero Jacome and Kalvin Chang and Darsh Agrawal and Keer Xu and Chao-Han Huck Yang and Jian Zhu and Shinji Watanabe and David R. Mortensen},
year={2026},
eprint={2601.14046},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2601.14046},
}
```