doc
Browse files
README.md
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language: fa
|
| 3 |
+
pretty_name: Farsi Youtube ASR Dataset
|
| 4 |
+
tags:
|
| 5 |
+
- Farsi
|
| 6 |
+
- Persian
|
| 7 |
+
- ASR
|
| 8 |
+
- youtube
|
| 9 |
+
license: cc0-1.0
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Farsi Youtube ASR Dataset
|
| 13 |
+
|
| 14 |
+
This dataset consists of over **385** hours of audio extracted from various YouTube videos in the Persian language.
|
| 15 |
+
|
| 16 |
+
## Note
|
| 17 |
+
This dataset contains raw, unvalidated transcriptions. Users are advised to:
|
| 18 |
+
1. Perform their own quality assessment
|
| 19 |
+
2. Create their own train/validation/test splits based on their specific needs
|
| 20 |
+
3. Validate a subset of the data if needed for their use case
|
| 21 |
+
|
| 22 |
+
## Dataset Description
|
| 23 |
+
|
| 24 |
+
The dataset includes Farsi content from various types of videos, including:
|
| 25 |
+
- Podcasts
|
| 26 |
+
- TV Shows
|
| 27 |
+
- Educational Content
|
| 28 |
+
- Interviews
|
| 29 |
+
- Documentaries
|
| 30 |
+
|
| 31 |
+
## Usage
|
| 32 |
+
<details>
|
| 33 |
+
|
| 34 |
+
Huggingface datasets library:
|
| 35 |
+
```python
|
| 36 |
+
from datasets import load_dataset
|
| 37 |
+
dataset = load_dataset('PerSets/fytasr', trust_remote_code=True)
|
| 38 |
+
```
|