There is an interesting recent post about the history of Google Music in China from an article by Michael Zhang in Global Entrepreneur Magazine. Some excerpts:
“It will play the right music without you having to give it any thought.” You get the music you want at the right time, for the right environment, and in the right mood. Well, it sounds unrealistic that one’s thoughts can actually control music. Yet like the famous science fiction writer Arthur C. Clarke said, “Any sufficiently advanced technology is indistinguishable from magic.” Revolutionary technologies like electrical power, airplanes and search engines have all changed the world in ways that were out of expectation. To some extent, Google’s second edition of their music search product, launched on March 30 by Hong Feng and his team, fulfilled the criteria in some way. When Google China’s president Li Kaifu [who by now quit Google] and executives from hundreds of record companies posed for a photograph at the media conference, people might have ignored the fact that this product transcended reality on at least two levels.
They’re calling Google Music, an extremely late entrant into the music search business, a revolutionary, magical technology that transcends reality? Are you kidding me? Have they forgotten about Moodlogic (1998)? Shazam (2000)? Pandora (2000 or 2004, depending on whether you’re talking B2B or B2C)? Last.fm (2004)? Echo Nest (2005-ish?) How does Google Music China compare with the solutions provided by these other companies?
Google Music unprecedentedly enriched the way people find music. You can find a song through the name of artist, titles of the song, albums, or even a sentence of lyric, and you can also play the hottest songs from the charts. However, the most impressive breakthroughs are these two functions – one can have music recommendations according to difference of the tempo, tone, and timber; similar songs are recommended according to the timber of specific songs. Fresh experience it offers and the technical complexity in its realization makes it the most ambitious and imaginative work of Google after it entered China.
“Unprecedentedly”? Come on. It’s totally “precedented”. Last.fm has been doing it for years now. George Tzanatakis has been publishing Music IR papers on timbre-based search since at least 2002 (and talking with me and other folks about it since ISMIR 2001); , and has open-source code (Marsyas) that has been used in Google-hosted Summer of Code music projects. Brian Whitman (of Echo Nest) has been publishing about merging text- and audio-based music search it since at least 2002. And perhaps this is just sour grapes, but in early 2002 Google announced their first annual programming contest, explicitly and actively soliciting from the web community interesting ideas about searches one could do on the web: (italics mine)
Your mission is to write a program (most likely by adding code to the ripper) that does something interesting with the data, in such a way that it would scale to a web-sized collection of documents. Part of your job is to convince us of why your program is interesting and why it will scale; other than that, you’re free to implement whatever strikes your fancy.
I wrote back to the 2002 contest organizers and said that what struck my fancy was music information retrieval, i.e. musical documents found in those web pages. Any music found on those web pages could be mined, run through audio analysis to extract not only tempo, tone and timbre, but also rhythmic structures (based on an intelligent combination of metric analysis — which is one step above simple tempo analysis — and timbral cues, e.g. high hat vs. bass kick) and harmonic structures (evolving chord sequences, which was the main focus of my 2004 dissertation). After being processed and indexed, you could then find similar pieces of music based on these musical audio structures. Is that not interesting? I thought so.
Google wrote back and said no. Don’t even bother entering the contest. We won’t accept a submission that deals with that type of data (music) extracted from web pages, because we’re concerned about violating copyright.
Well, let’s turn it around then, I replied. Let’s use our musical analysis to do song identification, so as to be able to filter out from web search results any pages with copyrighted music on them, thus preserving the integrity of Google’s search results and indemnify Google against the risk of getting sued. The technology existed at that point; I knew it was possible. All it took was the proper contextual application. If copyright concerns were such a problem, then let’s address that issue. Not that I had any illusions that such an idea would automatically win the programming contest. Only that I would be allowed to enter the contest, to compete. As the contest said, “You’re free to implement whatever strikes your fancy”.
Again, the answer was no; they would not accept such an entry.
Four years later (in 2006) Google bought YouTube, and immediately had a music audio copyright detection problem on its hands. Lawsuits began flying left and right. Instead of being prepared to detect / filter / identify copyrighted music, they were left scrambling and wasn’t until 2008 that (I believe) the system was finally put into place. And right around that same time (2006), according to the article above, they started working on content-based music similarity algorithms of the sort that I had proposed to them via the programming contest in 2002.
Now, I understand that large companies (or even small companies) have legal issues when it comes to accepting and implementing unsolicited ideas from the general population. They can’t always go with whatever they hear, and I appreciate that fact. But this wasn’t an unsolicited idea. This was a programming contest explicitly designed to elicit these sorts of ideas. And so to be told to not even enter, and then see articles like the above one years later, in which the same proposals are lauded as revolutionary and (especially) unprecedented, flies in the face of the history as I know it. I’m not saying that I was the only one at the time thinking about these sorts of issues. We started ISMIR in 2000, and there had already been two conferences by January 2002 in which a lot of us (almost 100 people attended each conference) were headily and excitedly discussing these brave new ideas. I am only saying that Google, or at least some people at Google, knew about these things for years. They had asked for these ideas from the community, and the community responded. There was a precedent.