Add dataset card
Browse files
README.md
CHANGED
@@ -34,6 +34,61 @@ dataset_info:
|
|
34 |
download_size: 1624762124
|
35 |
dataset_size: 2078070414.74
|
36 |
---
|
37 |
-
# Dataset Card for
|
38 |
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
34 |
download_size: 1624762124
|
35 |
dataset_size: 2078070414.74
|
36 |
---
|
37 |
+
# Dataset Card for NB Tale, module 1 and 2 (< 15 sec. segments)
|
38 |
|
39 |
+
## Dataset Description
|
40 |
+
|
41 |
+
- **Homepage:**
|
42 |
+
- **Repository:** <https://github.com/scribe-project/nodalida_2023_combined_training>
|
43 |
+
- **Paper:**
|
44 |
+
```
|
45 |
+
|
46 |
+
@inproceedings{
|
47 |
+
solberg2023improving,
|
48 |
+
title={Improving Generalization of Norwegian {ASR} with Limited Linguistic Resources},
|
49 |
+
author={Per Erik Solberg and Pablo Ortiz and Phoebe Parsons and Torbj{\o}rn Svendsen and Giampiero Salvi},
|
50 |
+
booktitle={The 24rd Nordic Conference on Computational Linguistics},
|
51 |
+
year={2023}
|
52 |
+
}
|
53 |
+
```
|
54 |
+
|
55 |
+
- **Point of Contact:** [Per Erik Solberg](mailto:per.solberg@nb.no)
|
56 |
+
|
57 |
+
### Dataset Summary
|
58 |
+
|
59 |
+
This is the version of the Bokmål segments of module 1 and 2 of NB Tale used for testing the models
|
60 |
+
in the paper *Improving Generalization of Norwegian ASR with Limited Linguistic Resources* presented at NoDaLiDa 2023.
|
61 |
+
It only contains segments of a length < 15 sec. This dataset contains both native and non-native speakers.
|
62 |
+
Speakers with `region` set to `foreign` were filtered out [when analyzing the data in the paper](https://github.com/scribe-project/nodalida_2023_combined_training/blob/main/analysis/analysis.ipynb).
|
63 |
+
|
64 |
+
### Languages
|
65 |
+
|
66 |
+
Norwegian Bokmål
|
67 |
+
|
68 |
+
|
69 |
+
## Dataset Creation
|
70 |
+
|
71 |
+
### Source Data
|
72 |
+
The full version of this dataset is found in [the repository of the Norwegian Language Bank](https://www.nb.no/sprakbanken/en/resource-catalogue/oai-nb-no-sbr-31/)
|
73 |
+
|
74 |
+
#### Initial Data Collection and Normalization
|
75 |
+
|
76 |
+
The data was retrieved using the [Spraakbanken downloader](https://pypi.org/project/spraakbanken-downloader/) and standardized
|
77 |
+
using the [combined dataset standardization scripts](https://github.com/scribe-project/asr-standardized-combined). Bokmål segments with a duration < 15 seconds were
|
78 |
+
extracted using [this code](https://github.com/scribe-project/nodalida_2023_combined_training/blob/main/make_datasets/make_nbtale_csvs.ipynb).
|
79 |
+
|
80 |
+
## Licensing Information
|
81 |
+
|
82 |
+
[CC0](https://creativecommons.org/share-your-work/public-domain/cc0/)
|
83 |
+
|
84 |
+
### Citation Information
|
85 |
+
```
|
86 |
+
|
87 |
+
@inproceedings{
|
88 |
+
solberg2023improving,
|
89 |
+
title={Improving Generalization of Norwegian {ASR} with Limited Linguistic Resources},
|
90 |
+
author={Per Erik Solberg and Pablo Ortiz and Phoebe Parsons and Torbj{\o}rn Svendsen and Giampiero Salvi},
|
91 |
+
booktitle={The 24rd Nordic Conference on Computational Linguistics},
|
92 |
+
year={2023}
|
93 |
+
}
|
94 |
+
```
|