CompSpoof V2: 250,000 Audio Samples for Component-Level Anti-Spoofing

CompSpoof V2 contains over 250,000 audio samples totaling approximately 283 hours, developed by XuepingZhang for component-level anti-spoofing research. The data simulates real-world acoustic scenarios where speech, environmental sounds, or both components are spoofed, with each sample provided at multiple sampling rates.

Use Cases

• Binary classification of genuine versus spoofed audio components
• Multi-label detection to distinguish between spoofed speech and spoofed environmental sounds
• Robustness testing of anti-spoofing models across different sampling rates

Strengths

• 250,000+ audio samples
• 283 hours of total duration
• Multiple sampling rates provided for each sample
• Fixed 4-second duration for uniform batch processing

Limitations

• Non-commercial license restriction (CC BY-NC 4.0)
• Geographic bias toward the United States as indicated by region tags

Provenance

Source

XuepingZhang via Hugging Face

Freshness

Last updated February 2026.

Geography

United States