hci-lab-dcug/bibletts-asante-twi-max29secs-total9hrs-sr22050

---
license: cc-by-sa-4.0
task_categories:
- text-to-speech
language:
- ak
- tw
size_categories:
- 1K<n<10K
---
# BibleTTS Asante Twi Dataset

## Dataset Information

This dataset is derived from the BibleTTS corpus, specifically focusing on Asante Twi speech data. The original BibleTTS is a large, high-fidelity, multilingual, and uniquely African speech corpus.

- **Total Duration**: {total_hours:.2f} hours ({total_hours*60:.1f} minutes)
- **Number of Files**: {file_count:,}
- **Sample Rate**: {sample_rate:,} Hz
- **Max File Duration**: {max_duration:.1f} seconds
- **Format**: WAV files with corresponding transcriptions
- **Language**: Asante Twi (tw)

## Dataset Structure

```
bible-tts-dataset/
├── wavs/
│ ├── ABC/
│ │ └── batch0-99/
│ │ ├── ABC_file1.wav
│ │ ├── ABC_file2.wav
│ │ └── ...
│ ├── DEF/
│ │ └── batch0-99/
│ │ └── ...
│ └── ...
└── dataset.jsonl
```

## Usage

Each entry in `dataset.jsonl` contains:
- `filename`: Relative path to the audio file
- `duration`: Duration in seconds
- `transcription`: Text transcription of the audio
- `speaker`: Speaker ID (3-letter code)

## Loading the Dataset

```python
import json
import librosa

# Load dataset metadata
with open('dataset.jsonl', 'r', encoding='utf-8') as f:
dataset = [json.loads(line) for line in f]

# Load an audio file
audio_path = dataset[0]['filename']
audio, sr = librosa.load(audio_path, sr={sample_rate})
transcription = dataset[0]['transcription']
speaker = dataset[0]['speaker']
```

## Statistics

- **Total Duration**: {total_hours:.2f} hours
- **Average File Duration**: {(total_hours * 3600 / file_count):.1f} seconds
- **Files**: {file_count:,} audio files
- **Multiple speakers**: Selected from train/dev/test splits

## Citation

If you use this dataset, please cite the original BibleTTS paper:

```bibtex
@inproceedings{{meyer2022bibletts,
title={{BibleTTS: a large, high-fidelity, multilingual, and uniquely African speech corpus}},
author={{Josh Meyer and David Adelani and Edresson Casanova and Alp {{\\"O}}ktem and Daniel Whitenack and Julian Weber and Salomon Kabongo Kabenamualu and Elizabeth Salesky and Iroro Orife and Colin Leong and Perez Ogayo and Chris Chinenye Emezue and Jonathan Mukiibi and Salomey Osei and Apelete Agbolo and Victor Akinode and Bernard Opoku and Olanrewaju Samuel and Jesujoba Alabi and Shamsuddeen Hassan Muhammad}},
booktitle={{Interspeech}},
publisher = {{{{ISCA}}}},
year={{2022}},
url={{https://arxiv.org/pdf/2207.03546.pdf}}
}}
```

## Dataset Origin

This dataset is a processed subset of the original BibleTTS corpus, specifically containing Asante Twi speech data. The original audio files were recorded at 48kHz and have been downsampled to 22.05kHz for this distribution.

## License

Please refer to the original BibleTTS licensing terms for usage permissions.