README.md

---
dataset_info:
  features:
  - name: speaker_id
    dtype: int64
  - name: sentence
    dtype: string
  - name: audio
    dtype:
      audio:
        sampling_rate: 16000
  - name: label
    dtype: string
  - name: gender
    dtype: string
  - name: tts
    dtype: string
  splits:
  - name: train
    num_bytes: 5004898198.44
    num_examples: 11130
  - name: validation
    num_bytes: 1402111399.43
    num_examples: 3181
  - name: test
    num_bytes: 695119245.014
    num_examples: 1591
  - name: fishaudio
    num_bytes: 360425141.0
    num_examples: 1000
  - name: xtts
    num_bytes: 217734249.0
    num_examples: 1000
  - name: mms
    num_bytes: 300196557.0
    num_examples: 1000
  - name: speecht5
    num_bytes: 184191717.0
    num_examples: 1000
  download_size: 7132788149
  dataset_size: 8164676506.884
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
  - split: fishaudio
    path: data/fishaudio-*
  - split: xtts
    path: data/xtts-*
  - split: mms
    path: data/mms-*
  - split: speecht5
    path: data/speecht5-*
---

```python
from datasets import load_dataset

dataset = load_dataset("elsayedissa/AFAD-MSA")
```


```python
DatasetDict({
    train: Dataset({
        features: ['speaker_id', 'sentence', 'audio', 'label', 'gender', 'tts'],
        num_rows: 11130
    })
    validation: Dataset({
        features: ['speaker_id', 'sentence', 'audio', 'label', 'gender', 'tts'],
        num_rows: 3181
    })
    test: Dataset({
        features: ['speaker_id', 'sentence', 'audio', 'label', 'gender', 'tts'],
        num_rows: 1591
    })
    fishaudio: Dataset({
        features: ['speaker_id', 'sentence', 'audio', 'label', 'gender', 'tts'],
        num_rows: 1000
    })
    xtts: Dataset({
        features: ['speaker_id', 'sentence', 'audio', 'label', 'gender', 'tts'],
        num_rows: 1000
    })
    mms: Dataset({
        features: ['speaker_id', 'sentence', 'audio', 'label', 'gender', 'tts'],
        num_rows: 1000
    })
    speecht5: Dataset({
        features: ['speaker_id', 'sentence', 'audio', 'label', 'gender', 'tts'],
        num_rows: 1000
    })
})

```

```
@article{issa2026afad,
  title={AFAD-MSA: Dataset and Models for Arabic Fake Audio Detection},
  author={Issa, Elsayed},
  journal={Computation},
  volume={14},
  number={1},
  pages={20},
  year={2026},
  publisher={MDPI}
}
```