In less than 24 hours, I’ve learned about two different ambient audio sampling and matching projects that I had never heard of. They sound a little like magic but audio fingerprints even, or especially, very small ones are very information rich and unique in many ways. So samplers beware.
The first is immi.com or Integrated Media Measurement Inc a market research firm that gives you a cell phone, if you are one of their “panelists” The phone then takes small random audio samples of the ambient audio in the phone’s reach, packs it into a fingerprint and sends it up to a database to be matched. Their How It Works section explains the process from panel selection to data analysis. Their reported panelist retention is as amazing as their technologies.
Now TechCrunch reports that Google Research is working in the same area. The Google Research papers are more about fingerprint creation and matching (the real magic here as audio is so wonderfully compressable if you don’t mind loss and if you are after a finerprint loss is fine).
A bit from TechCrunch here:
Google Research team members Michele Covell and Shumeet Baluja along with Michael Fink of the Hebrew University of Jerusalem’s Center for Neural Computation were given the best paper award for their report on the system at the the Euro ITV (interactive television) conference last week. (“Social- and Interactive-Television Applications Based on Real-Time Ambient-Audio Identification” 10 pg PDF, see also the Google Research blog post on the paper.)
But unlike IMMI.com, Google is sampling from “your home computer” instead of a special cell phone and it’s unclear how the matched information will be used, that is there are a lot of suggestions that don’t make sense to me.