MultiAPI-Spoof / README.md
jhsdfbsdjfu's picture
Update README.md
cdbb78b verified
metadata
license: cc-by-nc-4.0
extra_gated_heading: Acknowledge license to accept the repository
extra_gated_description: >
  The dataset is released under [CC BY-NC
  4.0](https://creativecommons.org/licenses/by-nc/4.0/deed.en) license. Users
  must comply with the license terms. The authors do **not** claim ownership of
  the original audio content.
extra_gated_button_content: Acknowledge license

MultiAPI Spoof: Multi-Source Audio Anti-Spoofing Dataset

Introduction

MultiAPI-Spoof is a multi-source audio anti-spoofing dataset that contains approximately 230 hours of spoofed audio. It includes synthetic audio generated by commercial TTS services, open-source models, and Chinese TTS websites. An equal amount of bonafide speech from CommonVoice is included for a 1:1 balance between genuine and spoofed samples. This dataset is designed to support research and model training for audio anti-spoofing.


Spoofed Audio Data Sources

Our new dataset, MultiAPI Spoof, contains speech samples generated from a variety of API sources, including:

  1. Commercial TTS APIs – speech generated by commercial services.
  2. Open-Source Model Generation – speech generated by open-source models.
  3. TTS Websites – speech on TTS web platforms.

The dataset is organized into 30 API, labeled A0–A29, with each group corresponding to a unique speech generation API source. The duration of speech in each API ranges from 0.2 to 12 hours.


Dataset Split

API NO. train dev eval
A0-A20 70% 10% 20%
A21-A23 / 100% /
A24-A29 / / 100%

Metadata

The dataset includes three metadata files: MultiAPI_train.txt, MultiAPI_dev.txt, and MultiAPI_eval.txt.

Each line has four fields:

audio_path     api     class_label
XXX.mp3        A0      spoofed
XXX.mp3        -       bonafide

📖 Citation

If you use this code or dataset, please cite:

Anonymous

📧 Contact

Anonymous


🔏 License

The dataset is released under CC BY-NC 4.0 license. Users must comply with the license terms. The authors do not claim ownership of the original audio content.