Re: Server side speech

At 7:24 PM -0400 4/25/07, Justin Frim wrote:

Perhaps lightly off-topic, I had to make a quick audio CAPTCHA to complement a visual one for a web site. I was thinking of having a server-side TTS system, but that just became too big a can of worms for the size of the project. I also had the restriction that the hosting service provider refused to install any machine-code executables or compile any programs to be installed on their servers. So instead I used a collection of 95 sound clips, each one being a recording of an ASCII character spoken out loud, with the file named as the 2-digit hex code of the respective character. All the PHP script had to do was concatenate the audio data portion of the necessary files, then create a new header and send the result as a single audio file to the user-agent. Worked like a charm. (I should also mention the audio clips were in IMA-ADPCM format, so a simple concatenation like that didn't break anything.)

I did a similar thing with this: http://sperling.com/examples/captcha/ After I write it up, I will be providing the code to the public. I simply used mp3 files (my own voice) and made all files the same size. That way, to create a key, I simply loaded the header that contained the length and then a set number of other files were appended to that -- the length was always the same. However, I did run my audio captcha by a couple dozen visually impaired testers to fix any problems that they might have -- it's interesting to see what they see. Cheers, tedd -- ------- http://sperling.com http://ancientstones.com http://earthstones.com