I guess they have to use some sort of obfuscation in order to prevent spam bots from using text-to-speech software to get around the mechanism, but that is REALLY confusing. My guess is that the audio characters don't necessarily match up with the ones displayed in the image, but all I could make out was A (S/F) M