Last month, Google announced the new no-captcha reCAPTCHA that is supposedly more accurate and better at preventing spams. We’ll see how this goes.
In the mean time, plenty of websites that employ Google’s reCAPTCHA still use the old version like this:
The problem with this reCAPTCHA is that it fundamentally doesn’t work with screen readers (among other things, like forcing you crossed your eyes trying to figure out each character in the string.) Some people pointed out that reCAPTCHA offers the sound version (see that little red speaker?) that should mitigate the problem.
Here’s the link to sound version of a Google reCAPTCHA: https://dl.dropboxusercontent.com/u/9074989/google-recaptcha-audio.mp3
This example was taken from the PubMed website and happened to be set as a string of numbers.
p.s. what is this a about PubMed using inaccessible reCAPTCHA? There are other ways to employ non-captcha security techniques without using that kind solution. :-/
p.p.s. In case you’re curious, I could not decipher two out of the eleven (if I counted it correctly) numbers said in that recording.