chiprook
← AI
AISeptember 18, 2026, 10:27

Study: Audio Processing Makes Real Speech Look Synthetic to Detectors

A researcher tested 13 synthetic-speech detectors on 47 real recordings and found normal audio processing (codec, denoise, EQ) makes them flag genuine speech as generated. A codec pass shifted 9 of 13 detectors, and 1984's Griffin-Lim shifted 12 of 13.

Study: Audio Processing Makes Real Speech Look Synthetic to Detectors
Read next
AI

OpenAI and Anthropic Negotiated Mutual AI Stress-Testing Deal

AI

AI tag team Jev and Astra beat Minecraft in 8m43s for under $1

AI

OpenAI details six real model misalignment incidents

AI

Space Daily Published AI Slop Under Fake NASA and ESA Expert Bylines