I'm currently working on a CAPTCHA plugin for DokuWiki and thought about providing audio output for users not able to see the image. This is pretty simple for CAPTCHAs – there is no need for complicated speech synthesis because you only need recordings of the 26 possible letters. But you need a way of joining those recordings on the fly…

I looked around the web for a way to concatenate multiple wave files with PHP and found an article by Phillip Perkins called Create an audio stitching tool in PHP. But at a closer look his example code was overly complicated for my simple task. I then stumbled about a very short forum thread which together with a simple format specification led to the following function:

function joinwavs($wavs){
    $fields = join('/',array( 'H8ChunkID', 'VChunkSize', 'H8Format',
                              'H8Subchunk1ID', 'VSubchunk1Size',
                              'vAudioFormat', 'vNumChannels', 'VSampleRate',
                              'VByteRate', 'vBlockAlign', 'vBitsPerSample' ));
    $data = '';
    foreach($wavs as $wav){
        $fp     = fopen($wav,'rb');
        $header = fread($fp,36);
        $info   = unpack($fields,$header);
        // read optional extra stuff
        if($info['Subchunk1Size'] > 16){
            $header .= fread($fp,($info['Subchunk1Size']-16));
        }
        // read SubChunk2ID
        $header .= fread($fp,4);
        // read Subchunk2Size
        $size  = unpack('vsize',fread($fp, 4));
        $size  = $size['size'];
        // read data
        $data .= fread($fp,$size);
    }
    return $header.pack('V',strlen($data)).$data;
}

Just fill it with a bunch of .wav files and it will return a new joined file. Send it with a audio/x-wav header and you're done.

What I need now are the audio files. I found some on the web but I can't figure out who owns them and what the license is… If anyone can help me out, mail me or leave a comment.

Similar posts:

• Using SQLite as Vector Store in PHP
• Weekend Project: indieblog.page
• Regular Expression to match Multi-Byte