1. Need a flash/flex type component that can take a recording or soundbyte from user on webpage(length of recording can be adjusted), convert the audio into wav or some small audio format. There should be an option for input audio as (actual audio) or (sung, hummed, tapped, etc audio).
2. The converted/recorded audio would then need to be compared to similar formats recognizing samples, by pitch, tempo variation, speech content and/or location of pauses(by melody) either in a database on the server or just the server itself. The actual audio would require more of an exact match.
3. The closest matches would then be returned to the user.
The component would be similar to that of [url removed, login to view], [url removed, login to view], or shazam.com. I know searching for music like this is complex to the programmer so fine details can be worked out. Thanks.