But how do we know that your not fake? So you see it is going to take a few reputable people running a test to give us an answer we can believe.
I downloaded this twitter video https://x.com/MSNBC/status/1815506040902832223
using this site https://twittervideodownloader.com
I extracted a clip from the middle of the Biden “call” and converted to audio only on this site https://biteable.com/tools/extract-audio-from-video/
I ran the evaluation on the clip here: https://elevenlabs.io/ai-speech-classifier
You can try it too if you want to.
I had my son run the same test and it also said it was 2% likely to be AI-generated by them. I asked my son what that meant. He said either something has changed with the recording that was submitted for evaluation, something changed with the AI site, or the guy posting on X was lying.
So who knows.
First we did it with the first minute of the clip from the CNN video and it said 2% and then we found the exact segment that the guy on X used and ran it.
I had the same questions that you had, which is why I had my son run it. I’m not sure how the guy on X could have faked it. And I’m not sure how the AI site could fake their results. To find out I suppose a person would have to do the same test on a sample known to have been generated by them, and see if it detects that it was done by them. Maybe I could have my son try that.