Datasets:

Modalities:
Text
Formats:
json
ArXiv:
Libraries:
Datasets
Dask
License:
File size: 1,416 Bytes
fcb5f8f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d2eb8ee
fcb5f8f
 
 
aea45a5
fcb5f8f
 
d57b1e8
fcb5f8f
 
 
 
 
 
 
925cded
fcb5f8f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
---
language:
- en
- de
- fr
- es
license: "odc-by"
tags:
- phonemization
- g2p
- multilingual
- olaph
pretty_name: "OLaPh Phonemization Dataset"
dataset_info:
  dataset_name: olaph-data
  version: "1.0.0"
  created_by: IISYS Hof University
---

# OLaPh Phonemization Dataset

## Dataset Summary
This dataset was created with the [OLaPh framework](https://github.com/iisys-hof/olaph) and used to train the [OLaPh grapheme-to-phoneme model](https://huggingface.co/iisys-hof/olaph). It contains multilingual text from the FineWeb datasets, automatically phonemized into text–phoneme pairs for English, German, French, and Spanish.

## Source Data
Text was taken from:
- [HuggingFaceFW/fineweb](https://huggingface.co/datasets/HuggingFaceFW/fineweb)
- [HuggingFaceFW/fineweb-2](https://huggingface.co/datasets/HuggingFaceFW/fineweb-2)

## Phonemization
All text was processed and phonemized automatically using the [**OLaPh framework**](https://github.com/iisys-hof/olaph).

## Intended Use
This dataset is intended for research and development of phonemization models, and related speech or linguistic tasks.


### Citation
```bibtex
@misc{wirth2025olaphoptimallanguagephonemizer,
      title={OLaPh: Optimal Language Phonemizer},
      author={Johannes Wirth},
      year={2025},
      eprint={2509.20086},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2509.20086},
}
```