A curated list of audio attacks, defenses, privacy, watermarking, and trustworthy audio AI.
| Category | Brief scope |
|---|---|
| Audio Deepfake, Voice Cloning, and Spoofing | Deepfake speech, TTS/VC misuse, spoofing, detection, tracing, protection. |
| Voice Privacy, Anonymization, and Speaker Protection | Anonymization, identity leakage, unlearning, private speech representations. |
| Watermarking, Provenance, and Data Rights | Audio watermarking, copyright, provenance, generated-audio authenticity. |
| ASR and Speech Translation Security | Attacks, defenses, privacy, and robustness for ASR and speech translation. |
| Audio-Language Model Safety | Jailbreaks, prompt injection, guardrails, benchmarks, hallucination, bias. |
| Side Channels and Physical Eavesdropping | Acoustic, vibration, sensor, meeting-audio, fiber, and device side channels. |
| Music and Singing Voice Security | AI covers, singing voice conversion, generated music detection, protection. |
| Voice Authentication and Biometrics | Voice liveness, speaker verification security, anti-spoofing, audio authentication. |
Current focus: 2025 and public 2026 top-venue papers from security, AI, NLP, and speech conferences.
Category files are split into broad sections: Attack, Defense, Benchmark & Measurement, and Other. Tables are sorted newest first.
The list favors main-conference/full research papers and strong arXiv preprints; short, Findings, workshop, tooling-only, and position papers are excluded by default.
Pull requests are welcome. Please include the title, authors, venue, year, official link, primary category, section, tags, and a short neutral security/privacy summary.
See Contributing for the preferred format.