Datasets:
Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- ha
|
| 4 |
+
license: cc-by-4.0
|
| 5 |
+
task_categories:
|
| 6 |
+
- automatic-speech-recognition
|
| 7 |
+
pretty_name: Lugha – Hausa Speech Dataset
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Lugha – Hausa Speech Dataset
|
| 11 |
+
|
| 12 |
+
Crowd-sourced Hausa voice recordings collected via the Lugha mobile app.
|
| 13 |
+
|
| 14 |
+
## Structure
|
| 15 |
+
|
| 16 |
+
```
|
| 17 |
+
audio/sample_XXXX.m4a – raw audio (m4a)
|
| 18 |
+
metadata.jsonl – one JSON object per recording
|
| 19 |
+
```
|
| 20 |
+
|
| 21 |
+
## Fields
|
| 22 |
+
|
| 23 |
+
| field | description |
|
| 24 |
+
|-------|-------------|
|
| 25 |
+
| audio | relative path to the audio file |
|
| 26 |
+
| text | prompt sentence read by the speaker |
|
| 27 |
+
| language | spoken language |
|
| 28 |
+
| state | Nigerian state of the speaker |
|
| 29 |
+
| lga | Local Government Area |
|
| 30 |
+
| accent | self-reported accent |
|
| 31 |
+
| age_range | speaker age bracket |
|
| 32 |
+
| gender | speaker gender |
|