Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -54,4 +54,47 @@ language:
|
|
54 |
- cy
|
55 |
size_categories:
|
56 |
- 10K<n<100K
|
57 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
54 |
- cy
|
55 |
size_categories:
|
56 |
- 10K<n<100K
|
57 |
+
---
|
58 |
+
[See below for English](#Voices-of-Arfor)
|
59 |
+
|
60 |
+
# Lleisiau Arfor
|
61 |
+
|
62 |
+
dod yn fuan...
|
63 |
+
|
64 |
+
|
65 |
+
# Voices of Arfor
|
66 |
+
|
67 |
+
This dataset was created at Cymen as part of a project funded by [Arfor](https://www.rhaglenarfor.cymru/index.en.html) in collaboration with the [Language Technologies Unit](https://huggingface.co/techiaith) at Bangor University.
|
68 |
+
|
69 |
+
The goal of the project was to collect a large amount of high quality Welsh speech data and their corresponding transcriptions with a particular focus on informal, conversational and spontaneous speech from the Arfor area. The resulting dataset will then be used to improve Welsh speech recognition technology and ensure the availability of the Welsh language in the latest technological advancements.
|
70 |
+
|
71 |
+
To achieve this, the project officer obtained permission to use already existing podcasts and record meetings, public events and conversations between volunteers. All of the data has been anonymised and is being released under an open (CC0) license.
|
72 |
+
|
73 |
+
The transcription style loosely follows the guidelines of the Language Technologies Unit’s [Banc Trawsgrifiadau](https://huggingface.co/datasets/techiaith/banc-trawsgrifiadau-bangor), particularly, in punctuation and data formatting while it diverges particularly with regards to formalising spelling and improving readability.
|
74 |
+
|
75 |
+
The dataset consists of three splits “test”, “train” and “dev” as well as a “clean” version for each of those data splits. The “train” split contains 80% of the data while “test” and “dev” contain 10% each. In the clean versions, all linguistic annotations and special characters have been removed to minimise the need for data formatting although the fully annotated data splits can still be used to customise the dataset.
|
76 |
+
|
77 |
+
Annotations include information such as:
|
78 |
+
- English or other foreign language words and segments indicated by asterisks, for example *spooky*
|
79 |
+
- two different options for transcribing numbers separated by the pipe character | and surrounded by curly brackets, for example {dau|2}
|
80 |
+
- paralinguistic sounds, such as \<chwerthin>
|
81 |
+
- filler words and sounds, such as “yy” and “yym”
|
82 |
+
|
83 |
+
This is an example of the data:
|
84 |
+
|
85 |
+
```
|
86 |
+
path sentence accents language
|
87 |
+
file9476.wav Galwad yn sydyn i bawb yn y gynulleidfa ac i'ch ffrindiau ac i yy eich teulu. Gogledd Ddwyrain cy
|
88 |
+
file1726.wav Trwy'r ymgyrch *Black Lives Matter* wnaeth bobl ifanc, a lot o bobl ifanc sylwi... Gogledd Orllewin cy
|
89 |
+
|
90 |
+
```
|
91 |
+
The dataset consits of four columns: path, sentence, accent and language.
|
92 |
+
|
93 |
+
| Column| Description |
|
94 |
+
| ------ | ------ |
|
95 |
+
| `path`| The path or file name in the 'clips' folder|
|
96 |
+
| `sentence`| The transcription|
|
97 |
+
| `accent`| The accent of the speaker, either: Gogledd Orllewin, Gogledd Ddwyrain, Canolbarth, De Ddwyrain, De Orllewin, Patagonia|
|
98 |
+
| `language`| The language of the entire segment, either English, if all of the words are English, or Welsh, if at least one word in the segment is Welsh|
|
99 |
+
|
100 |
+
If you have any questions about this dataset please contact [email protected]
|