The Internet has changed the way we listen to music. Now, scientists in Europe are working on a smarter way to look for it.
Smarter music browsing. I'm Bob Hirshon and this is Science Update.
If you find a song you like online, it can be tough to find others like it. But a powerful new music browser could help. It’s called SIMAC, and it’s spearheaded by Xavier Serra of the Pompeu Fabrea University in Barcelona, Spain. Rather than searching for titles and text descriptions, SIMAC analyzes the musical properties of a digital audio track.
So we can search for songs that have either a general concept of similarity, or a specific one, like search for similar rhythms. Or search for similar key signatures. Or search for songs that have the same style.
This could please not only consumers looking for fresh tunes, but also independent musicians, who could get worldwide exposure without building a big name. I'm Bob Hirshon, for AAAS, the science society.
Making Sense of the Research
Almost since music has been sold on the Internet, websites have been trying to make recommendations to their music customers. For example, click on Ludacris Presents Disturbing Tha Peace on one popular website, and it will also recommend albums by Jamie Foxx, Mary J. Blige, and Juelz Santana. If you like Ludacris, maybe you would like these albums, or maybe you wouldn’t. Recommendations can never be 100 percent accurate, even from your best friends.
The trouble with Internet music recommendation programs (and recommendations for books, DVDs, or anything else that’s a matter of personal taste) is that they’re determined indirectly. Usually, they’re predictions based on what you yourself have downloaded in the past, and what millions of other people who downloaded the same song have also downloaded. A lot of math goes into making these predictions, and they’re certainly more accurate than random guessing. But when a friend tells you about a song you might like, your friend knows what the song sounds like, and what the music you like sounds like. Most music websites, on the other hand, don’t know anything about the music itself; they just know about the choices people make.
If you want to browse for music yourself, you’re also limited. Almost all music search engines are text-based—meaning you can search only by words that you can type into a computer. Yes, the word might be the name of an artist or song, a style of music, or even descriptive words like “jazzy” or “dark,” but if you can’t quite put what you’re looking for into words, you can waste a lot of time.
SIMAC is an attempt to make music browsing a little more like asking a friend, or at least a helpful clerk at a CD store. Unlike most music browsers, it can analyze characteristics of the music itself, including the tempo (how fast or slow it is), the key signature (which can affect the mood of the music), and the rhythm (be it hip-hop, country, garage rock, or Latin music). If you know a bit about music, you can ask SIMAC to search specifically along any of these dimensions—for example, to find songs with a rhythm like Ne-Yo’s “So Sick”—or, to just look for songs that are generally similar.
How can this help independent musicians? Again, think about the differences between SIMAC and a conventional search engine. A website that makes recommendations based on what other people buy or download is usually going to recommend the big sellers (look back to the Ludacris-related recommendations in the first paragraph). As they get more recommendations, those artists get even more popular, leaving the rest of the pack in the dust. But a system that searches by a song’s musical properties can give an equal shot to everything in its database, including a demo made in your basement. International music that doesn’t have a title in our alphabet can also compete, even though these songs wouldn’t come up in a normal text-based search.
Of course, even SIMAC won’t be perfect. What sounds similar to a computer program may not sound similar to you. And even if it does, you may or may not like what you find. But it could make music browsing more versatile and efficient for consumers, and fair for musicians themselves.
Now try and answer these questions:
- What are the limitations of most music browsers and search engines?
- How is SIMAC different?
- What are the limitations of SIMAC?
- Do you agree that SIMAC might help independent musicians get their music to a wider audience? Why or why not?
You may want to check out the January 27, 2006 Science Update Podcast to hear this Science Update and the other programs for that week. This podcast's topics include: A better way to browse music; sexual orientation in the brain; a great locust migration; the tectonic future of California; why the desert is an Amazon.
To explore the legal aspects of digitized music, see the New York Times Learning Network’s lesson Bootleg Bytes. In this lesson, students conduct background research to learn about the positions of different groups involved in the digital piracy debate, then participate in a fishbowl discussion that seeks to find a consensus on the issue.
Read the How Stuff Works article How Internet Search Engines Work for background on traditional search engines.