Google music search 8 years late

Eight years ago Nick Whyte and our team at AltaVista invented multi-media search. We were the first with Image (picture) Search, Video Search, and Music Search. Seven years ago I joined the team at Napster where we invented P2P music search and file sharing. Pioneers sometimes win fame, but rarely earn fortunes. In this case, the not so fast "fast follower" Google will probably make the fortune.

C/Net broke the story Google whistles a new tune. Funny, I went to Google this morning and couldn't find it. It only shows up as a secondary option after returning results of a search. For example, if you enter "Vanessa Williams" in the regular search box Google will return a results page as usual. One of the first results will have a link that says "More music results for Vanessa Williams". This link spawns a new page with music related results and a new "Search Music" button.

All that multi-media search technology we developed at AltaVista is now the property of Yahoo. Yahoo acquired AltaVista several years ago and many of the brilliant AV engineers are still there. In fact, the Image, Video, and Music search technology is still working on the Yahoo site today.

Incidentally, Nick Whyte now works with the MSN Search group at Microsoft. MSN Search already has Image Search and could add Music Search at anytime. Gee, could Video Search be next?

Both AltaVista and Napster were ahead of their time. The technology was great but the business model hadn't developed and the market wasn't quite ready. In both cases there were several strategic decision blunders as well. I wrote blog articles on my time at AltaVista and Napster that detail the story and lessons learned.

Maybe Google isn't 8 years late...maybe AltaVista was 8 years early. We had the technology at AltaVista and we tried to make ecommerce deals with Amazon and CDNow for music, and Corbis and Getty Images for Images, but we were just too early for the market. Eight years later, maybe the time is right.

Comments

[eduo]

This would be a nice bit of history if it wasn't peppered by ill feelings from someone these days finding fortune in something that didn't prove very profitable when done before its time. I would've rather read about the actual history and roadblocks that altavista found implementing this (however unsuccessful or flawed it could've turned out to be) and a theory on why it wasn't a hit then (sorry, "Google is popular and that's why it'll be popular this time around" just doesn't cut it and sounds awfully unprofessional, because it WAS NOT successful when tried the first time around and Google wasn't around, so the reasons behind it not being a hit are not related to google, and it's not like Altavista wasn't inmensely popular back then).
As it stands right now, this entry is misleading and sounds like you're burned of not having had the same attention back when you did what you feel was the same back 8 years ago.
My take? Altavista was very popular and very innovative but was going ahead too fast and giving a lot of options to a public that wasn't interested or didn't have the technical means to be (broadband wasn't as ubiquous), coupled with users getting more and more the feeling of Altavista being a corporation out to get their money (not that Google isn't, it's just not as obvious).
That is just my take, and the same opinions plus some other rather harsher ones apply to Yahoo! as well (and its demise from being the favorite to being generally despised and ignored)

[JG]

Don,
I feel your frustration. I'm in a similar situation, having done music information retrieval for the past six years. Google is definitely quite behind, and yet people still think that they innovate. It's certainly quite annoying, I agree.

But let me submit this to you: Even at Alta Vista, you weren't the first people to do image retrieval. You might have been the first commercial web search engine to implement it. But, at least according to Wikipedia, image retrieval itself has been around for 14 years, six years longer than you:

http://en.wikipedia.org/wiki/CBIR

In fact, not only have academics been doing image retrieval even before AltaVista existed, they have also been working on content-based image retrieval, where you are using actual properties of the photograph such as shape, for longer than AltaVista has been around.

I don't know the details of what you did, but my guess is that at AV you used the text of the filename, the text of the surrounding web page in which the image was located, and maybe, maybe some color histogram information to do your retrieval. Well, all that stuff existed long before AltaVista.

One thing to realize right now is that Google isn't really doing music retrieval yet. They are doing text retrieval, on a subset of data that is music-related. What I mean is that the actual matching is going on at a text-level, not at an actual music/audio signal level.

I would argue that it's not really music retrieval, unless you're dealing with music data. So I submit to you that you were not doing music retrieval, but neither is Google. The commmercial field in this area is still wide open.

However, once again, academics have been doing this for years now. I saw my first ever content-based music search paper in 1993. Another one in 1995. A handful more in 1997 and 1998. And today, you have a whole community working on the problem. A whole community that -doesn't- include Google.

http://www.music-ir.org/mirex2006/index.php/Audio_Music_Similarity_and_Retrieval