SparkAudio/voxbox

1---2license: cc-by-nc-sa-4.03language:4- zh5- en6tags:7- speech8- audio9pretty_name: voxbox10size_categories:11- 10M<n<100M12task_categories:13- text-to-speech14---15 16# VoxBox17 18This dataset is a curated collection of bilingual speech corpora annotated clean transcriptions and rich metadata incluing age, gender, and emotion.19 20## Dataset Structure21 22```bash23.24├── audios/25│   └── aishell-3/                      # Audio files (organised by sub-corpus)26│   └── ...27└── metadata/28    ├── aishell-3.jsonl29    ├── casia.jsonl30    ├── commonvoice_cn.jsonl31    ├── ...32    └── wenetspeech4tts.jsonl          # JSONL metadata files33```34 35Each JSONL file corresponds to a specific corpus and contains one metadata record per audio sample.36 37## Metadata Format38 39Each line in the *.jsonl files is a JSON object describing one audio sample. Below is a typical example:40 41```json42{43  "index": "VCTK_0000044280",44  "split": "train",45  "language": "en",46  "age": "Youth-Adult",47  "gender": "female",48  "emotion": "UNKNOWN",49  "pitch": 180.626,50  "pitch_std": 0.158,51  "speed": 4.2,52  "duration": 3.84,53  "speech_duration": 3.843,54  "syllable_num": 16,55  "text": "Clearly, the need for a personal loan is written in the stars.",56  "syllables": "K-L-IH1-R L-IY0 DH-AH0 N-IY1-D F-AO1 R-AH0 P-ER1 S-IH0 N-IH0-L L-OW1 N-IH1 Z-R-IH1 T-AH0 N-IH0-N DH-AH0-S T-AA1-R-Z",57  "wav_path": "vctk/VCTK_0000044280.flac"58}59 60```61The corresponding audio file is located inside the extracted .tar.gz archive.62 63**Key Fields:**64 65- `index`: Unique identifier for the audio sample.66- `split`: Train/test split.67- `language`: Language of the audio sample. Currently only English and Chinese are supported.68- `age`, `gender`, `emotion`: Speaker and utterance attributes69- `pitch`, `pitch_std`, `speed`: Acoustic features70- `duration`: Duration of the audio sample in seconds71- `speech_duration`: Duration of the speech in seconds by excluding silence in both ends.72- `syllable_num`: Number of syllables in the utterance73- `text`: Transcription of the utterance74- `syllables`: Syllable-level transcription75- `wav_path`: Path to the audio file76 77 78## 📚 More Information & Download Instructions79 80For detailed information about the dataset, including download scripts and usage instructions, please visit the official GitHub repository:81 82🔗 https://github.com/SparkAudio/VoxBox83 84 85## 📌 Licence & Attribution86 87Please refer to the original licenses of each sub-corpus. This dataset merely aggregates and annotates the metadata in a unified structure for research purposes.88 89 90## 📬 Citation91 92If you use this file or its associated data in your research, please consider citing:93 94```bibtex95@article{wang2025spark,96  title={Spark-tts: An efficient llm-based text-to-speech model with single-stream decoupled speech tokens},97  author={Wang, Xinsheng and Jiang, Mingqi and Ma, Ziyang and Zhang, Ziyu and Liu, Songxiang and Li, Linqin and Liang, Zheng and Zheng, Qixi and Wang, Rui and Feng, Xiaoqin and others},98  journal={arXiv preprint arXiv:2503.01710},99  year={2025}100}101```