Apparently, Capitol Police and the FBI are involved in investigations of an alleged recording of Roger Stone, in which he calls for the assassination of Democratic congressmen Eric Swalwell and/or Jerry Nadler. (Mediate published the audio clip last week.)
edition.cnn.com/…
Stone denies ever saying such things, and called the audio clip “AI manipulation.”
Digital audio processing has advanced significantly in recent years, and AI is surely gaining ground in that realm. (This is why so many actors and voiceover artists were expressing concern during the writers strike in Hollywood.) I am a musician/composer who uses digital audio production software, but have no experience with or expertise in AI development or programming, and no knowledge of the practical limits of detecting such technology in audio recordings.
But I do know that with virtually any digital product it is usually possible to conduct forensic analysis and look for irregularities in code or artifacts which indicate how the product was produced, so it seems like some kind of “genuine or fake/altered” determination should be possible in this case.
Stone has been one of the most damaging figures to our democracy for decades, and it would be deeply satisfying to see him go down. But I’m not sure if this will stick. I’m willing to consider the possibility that the Stone recording is fake. If it is, then it could be illustrative of just how precarious our hold on fact and truth is becoming, especially within the context of the upcoming election. If it is real, then the question becomes what are the possible legal consequences for Stone. If prosecution is possible, any proof of the authenticity of the recording would need to be absolutely concrete.
Are there any Kos Geeks here with relevant experience who might be able to offer some enlightenment on what investigators might be looking for and what methods they might employ to analyze the audio?