« FMC vendor update: FirstHand Technologies | Main | Enterprise VoIP hardware CPE 2Q07 »

A search engine for hummingbirds

Some six years back I heard John Chambers of Cisco on CNBC telling a reporter how Internet would change our lives. The first example he gave was how one would be able to listen to his song of choice on his car stereo instantly downloaded from the internet. That got me thinking how awkward it would be to search for a song while driving, then selecting out of the search results, and then some more clicks here and there. I am sure such an application would be banned by the traffic police everywhere.

What if instead you are able to sing your song of choice into the Internet connected car stereo and it matches a pitch pattern and gives you possible options? That is what the search technology should do: it should match the pitch pattern of your voice with the list of pitch patterns already cataloged in the engine. What is important is that you are doing it hands-free. And if ‘pitch pattern’ matching is the criteria then you do not even need to remember the lyrics. You can even hum a string of music if it is a latest song where you do not know the lyrics. That happens often times in my case. I hook on to a new song, but being a new release I have to guess lyrics and search for such a song on the Internet. It is frustrating.

This audio-to-audio search could also be useful for searching podcasts. Suppose I want to locate all those webinars that mention Nokia’s E61 dual mode handsets. I can speak the words ‘Webinar Nokia E61’ into the audio search interface and get the search results. Again, the pitch related to the words spoken can be analysed and matched. There is no limit in terms of the applications where this technology can potentially be used.

Having said that, there are several companies out there that are working on audio search solutions. However 99% of those efforts are focused on audio-to-text conversion and then searching by text. Even the podcast searching is focused on voice recognition which involves speech-to-text conversion at some point. That is a poor way to do podcast search.

In my primary school days we used to play you-hum-I-guess on evenings when we used to have power outages. I (really) looked for such an option on KaaZa several years later. I thought the technology might have caught up with the processing power of primary school kids. Apparently our best is still not good enough.

Post a comment

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)

TrackBack

TrackBack URL for this entry:
http://www.ilocus.com/cgi-bin/mt/mt-t.cgi/185