---
configs:
- config_name: default
  data_files:
  - split: train
    path: task_1770352558371/*.parquet
dataset_info:
  features:
  - name: audio
    dtype: binary
  - name: transcription
    dtype: string
  - name: file_name
    dtype: string
  splits:
  - name: train
    num_examples: 2450
task_categories:
- automatic-speech-recognition
- text-to-speech
language:
- vi
size_categories:
- 1K<n<10K
license: bsd
---

# Voidces

Audio dataset with transcriptions for voice training.

## Latest Upload: task_1770352558371

- **Samples**: 2450
- **Parquet files**: 5
- **ZIP file**: `task_1770352558371/dataset_audio.zip`
- **Metadata**: `task_1770352558371/metadata.json`

## Dataset Structure

Files are organized by task ID:
```
task_1770352558371/
├── train-00000-of-00005.parquet
├── train-00001-of-00005.parquet
├── ...
├── dataset_audio.zip
└── metadata.json
```

Each parquet file contains:
- **audio**: Binary audio data (WAV format)
- **transcription**: Text transcription
- **file_name**: Reference (format: audio/name_00001.wav)

The `metadata.json` file contains:
- Processing parameters
- Detailed segment information
- Summary statistics
- Timestamps and file sizes

## Usage

```python
from datasets import load_dataset
import json

# Load dataset
ds = load_dataset("Translsis/Voidces", data_files="task_1770352558371/*.parquet")

# Load metadata
import requests
metadata_url = "https://huggingface.co/datasets/Translsis/Voidces/resolve/main/task_1770352558371/metadata.json"
metadata = requests.get(metadata_url).json()

# Or download ZIP
# https://huggingface.co/datasets/Translsis/Voidces/resolve/main/task_1770352558371/dataset_audio.zip

# Access audio
import io
import soundfile as sf

sample = ds['train'][0]
audio_bytes = sample['audio']
audio_array, sr = sf.read(io.BytesIO(audio_bytes))
```

## Stats

- **Total samples**: 2450
- **Parquet files**: 5
- **Format**: WAV (binary bytes)

Created with automated pipeline: Whisper → YAMNet → BS-RoFormer