Additional Resources

Duquesne Law review on AI voice cloning and consent

The phoneme set used by most English TTS models

International challenge producing open-source anti-spoofing models

Wikipedia overview of audio deepfake technology

Reusable, imperceptible audio perturbations that hijack deployed voice agents into taking actions (IEEE S&P 2026)

Military telephone network that used DTMF keys A–D

California SB-1001 requiring bot disclosure

Recording laws by jurisdiction

CISA guidance on recognizing and defending against social engineering and phishing attacks

CISA guide to recognizing and reporting phishing

Detector error rising roughly 30x on synthesis models it hasn't seen before (arXiv)

Official text and resources for the EU AI Act

Flow-matching zero-shot TTS — MIT code, but CC-BY-NC (non-commercial) weights

FBI's portal for reporting internet crime including vishing and voice cloning scams

FCC ruling making AI-generated voice robocalls illegal

Consumer guide on robocalls and AI voices

Learn FFmpeg libav the hard way — from zero to hero

FTC consumer alert on how scammers use AI to enhance family emergency schemes

FTC guidance on preventing harms from voice cloning

FTC phishing scam prevention resources

Models verbally refuse while the tool call still executes — a blind spot unique to voice, where users only hear the refusal

Comprehensive introduction to training and fine-tuning audio models

Apache-2.0 LLM-decoder ASR model leading open English word error rate

How callback verification can fail and best practices for making it effective against BEC and vishing

U.S. Senate bill on call center operations

Security awareness training platform with vishing simulations and deepfake training modules

Audio-native end-of-turn detection, replacing transcript-based approaches

MIT-licensed audio watermarking — embed and detect entirely locally

Long-form multi-speaker TTS; Microsoft pulled the code after misuse and community forks preserved it

Open-weight TTS cloning from roughly 3 seconds — weights are CC-BY-NC (non-commercial)

Phishing for information technique documentation

Why families and coworkers need a safe word to defend against AI voice cloning scams

State-level AI legislation tracker

Computer crime statutes by state

Academic survey on neural speech synthesis (arXiv)

NIST definition and resources on social engineering

NIST digital identity guidelines covering authenticator types, knowledge-based authentication limitations, and multi-factor requirements

NIST guidelines on authentication including why voice biometrics alone are no longer sufficient

NIST guidelines on out-of-band authenticators and using separate communication channels for verification

High-throughput open ASR model covering 25 languages

The fundamental limit on digital audio sampling rates

Community benchmark ranking open speech recognition models by word error rate

Fully open (BSD-2) audio-native turn detection model — 8MB, roughly 12ms on CPU

Guide to email aliasing for privacy

Apache-2.0 speech-in/speech-out model with native function calling

Alibaba's open TTS suite (Apache-2.0) — 3-second voice cloning, VoiceDesign, and CustomVoice

Official hosted demo of Qwen3-TTS voice cloning and voice design

All Qwen3-TTS checkpoints — 0.6B/1.7B Base, CustomVoice, and VoiceDesign variants

The Qwen3-TTS paper — architecture and 3-second voice cloning results (arXiv)

Inaudible near-ultrasonic jailbreak prompts delivered through microphone nonlinearity (USENIX Security 2026)

Wikipedia overview of social engineering in security

Comprehensive FFmpeg tutorial and reference

How vector similarity search works

The newest Whisper weight (October 2024) — still the most recent release

Pre-converted GGML models for whisper.cpp

Fraud prevention best practices for wire transfers including dual control and voice verification procedures