Mango: 'Bing Audio' dynamically analyzes music for faster matching
One of the cooler features we're looking forward to in WP7.5 aka Mango is the ability to use Bing to ID songs playing on the radio, TV, or in bars (and then leveraging our Zune Pass for instant gratification). The service is called Bing Audio and looks to be quite killer.
Over on the Windows Phone Blog and Channel 9 today, they've been highlighting the development of Bing Audio, what it means, how it works and what makes it different from say, Shazam. In regards to the latter part, one main area of difference will be the speed.
In Shazam, you have to "sample" the music for 10-15 seconds and then it sends it to the cloud to be analyzed, hopefully returning a result. In Bing Audio, within 3 seconds the fingerprinting service begins to try to ID the song--it doesn't need the whole sample but theoretically can ID the song as soon as there's enough info to get the job done. They're also not sending the audio over the air, but rather just the "fingerprint" aka specific acoustic properties. That means data also will be kept to a minimum.
Overall though, it sounds like they took into consideration ambient noise (bars), speed, the database and as a result, they can even ID some songs based of of the live version, which is pretty impressive. Of course, until we test it in the real world, we'll have to wait and see.
On Channel 9, you can watch Brandon Watson interview Steve Cosman, who helped develop the feature for Windows Phone. With the technology developed for the IDing and the vast Zune library, this should be a pretty killer feature (and you won't have to worry about Shazam's paid-for "unlimited tags", natch).
Check out the video after the break (HTML5/Silverlight, sorry mobile users!)