README.md

---
license: cc-by-nc-sa-4.0
task_categories: ["automatic-speech-recognition", "text-to-speech"]
language: ["vi"]
pretty_name: FPT Open Speech Dataset (FOSD)
size_categories: ["10K<n<100K"]
dataset_info:
  features:
  - name: audio
    dtype: audio
  - name: transcription
    dtype: string
  splits:
  - name: train
    num_bytes: 684961355.008
    num_examples: 25917
  download_size: 819140462
  dataset_size: 684961355.008
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

# unofficial mirror of FPT Open Speech Dataset (FOSD)

released publicly in 2018 by FPT Corporation

100h, 26k samples

official link (dead): https://fpt.ai/fpt-open-speech-data/

mirror: https://data.mendeley.com/datasets/k9sxg2twv4/4

DOI: `10.17632/k9sxg2twv4.4`

pre-process:
- remove non-sense strings: `-N` `\r\n`
- remove 4 files because missing transcription:
  - `Set001_V0.1_008210.mp3`
  - `Set001_V0.1_010753.mp3`
  - `Set001_V0.1_011477.mp3`
  - `Set001_V0.1_011841.mp3`

usage with HuggingFace:
```python
# pip install -q "datasets[audio]"
from datasets import load_dataset
from torch.utils.data import DataLoader

dataset = load_dataset("doof-ferb/fpt_fosd", split="train", streaming=True)
dataset.set_format(type="torch", columns=["audio", "transcription"])
dataloader = DataLoader(dataset, batch_size=4)
```