#2246 Add audio captcha to FAS signup page
Closed: Fixed None Opened 13 years ago by sijis.

When creating a new account on FAS, there is a CAPTCHA that needs to be
filled into the form. It would be great if there was a blind user
friendly alternative through an audio output, informing about the present
characters.

See for complete message:
http://lists.fedoraproject.org/pipermail/websites/2010-June/008193.html


We got this enhancement via the webmaster@ address.

The only one I know of so far is recaptcha. We'd need to re-evaluate whether externally hosted, proprietary services are okay in fedora infrastructure. This would include things like google search, recaptcha, and third-party hit tracking. I think we've placed cloud hosting into a separate category and approved it with the caveat that we prefer open source to proprietary.

pingou had the idea to implement this ourselves by taking text, passing it to one of the text to speech programs and then sending the generated audio file to the user. Sounds basically doable but I'm glossing over many of the details. Suppose it depends if someone is interested.

Things that are available:

  • espeak is in el
  • python-ttsx has a ctypes interface to espeak
  • espeak -w foo.wav "12 + 3 =" is a way to get the audio file on the command line

Ok I have a working proof of concept, it generates the captcha and when asked generate the audio file (wav) which is then read, deleted and the content is returned to the user with the right html header.

I will look into python-ttsx to avoid having to make system calls.

In the future I think it would be great if we could add localisation (generate the audio file based on the language that the user is using in fas).

pingou thanks for your work. Is there any ETA for deployment. I think this is not of minor priority. I am espeak maintainer and I am receiving such requests from visually impaired people who would like to contribute to Fedora.

Well the proposed patch need to be reviewed first.

I would also like to try avoiding the system call by looking into pyttsx but I have no idea how this extension works, so maybe this could be a latter revision.

Suggestions welcome :-)

Ok so there is a new version of the patch.

Next to the captcha, there is now a button to open a small pop-up window which shows a page which plays the content of the captcha. This uses pyttsx so no more system call directly :-)

damn it! I should have tested using another machine earlier, when I use a different client to call FAS it's still the server who's speaking.

I do see an interest to spice up the server room but not quite for our purpose, so let's forget this version.

Added option to enable/disable the audio component of the captcha
0001-Implement-audio-captcha.3.patch

Code has been committed and will be in the next release, scheduled for January. Leaving open until we deploy it.

Thanks pingou!

Login to comment on this ticket.

Metadata