This article talks about audio fingerprinting technology that can verify the name of a song when it plays on an Internet server. The article explains the following: the creation, usage, advantages of and possibility for future development of the system.
Sometimes people have to wait in vain to get the title of a song when they listen to the tune on the radio. To resolve this inconvenience, nowadays some companies are offering a technology that can identify a title of a tune while it is playing. For instance, this system shows words like Take Five by the Dave Brubeck Quartet on an Internet radio or cellphone.
The technology can recognize not only names and artists within a huge range of recorded music, but also “different versions of a piece done by the performers, even when the differences are slight.”
For example, at the recent Consumer Electronics Show in Las Vegas, Royal Philips Electronics showed a model of an Internet radio that can verify the name of the band when the music was played, and also “can distinguish a version of a tune that it played at a concert in Verona, Italy, from the same tune recorded in Milan.”
A mathematician at Philips Research in Eindhoven, the Netherlands, Ton Kalker (the leader of the team that made the rapid identification system) said that the technology is sensitive enough to make distinctions that humans can’t.
This technology is called audio fingerprinting. It is based on the idea that every performance of a song has unique audio characteristics.
For instance, “a certain relationship of neighboring high and low notes over a minuscule slice of time.” Audio fingerprinting makes those relationships in numbers, and slight differences in code indicate a separate version of a song.
Dr. Richard Gooch, deputy director of technology at the International Federation of the Phonographic Industry (a trade organization based on London), explained more about this system: Audio fingerprinting operates by making a mathematical explanation of some of the unique characters of a song. The fingerprints are saved on a server. And when a tune is playing on Internet radio and the system needs to verify the song, this technology matches a small piece of the tune, represented in code, with the entire coded version of the song stored on the server.
According to Dr. Gooch, some companies use fingerprinting technology to distinguish, not only streaming content, but also traditional radio broadcasts and the contents of audio files.
Dr. Gooch also said that this system might help the recording industry to make a profit (by monitoring which songs are played and thus being able to collect royalties). His group and the Recording Industry Association of America are examining many audio fingerprinting systems for that.
Dr. Gooch emphasized that the technology can work even in very bad conditions such as “poor loudspeakers, highly compressed streaming files or broadcasters who speed up songs slightly to make room for commercials.” He mentioned that as long as people can hear a song (even in poor circumstances), “the systems can extract the description of unique characteristics in the song, quickly matching the description with the database to identify the track.”
According to him, since the early 80’s, this technology has been improving little by little, and has been used in the business field. For instance, it has been used in the music industry to verify broadcast performances and then give royalties to “right holders.”
Dr. Gooch said “Audio fingerprinting is accurate, robust and runs in a sensible amount of time. It really works.”
He also mentioned that consumer applications of the technology are probably needed because of the popularity of digital music.
According to the article, Shazam Entertainment, based in London, already gives consumers an audio fingerprinting service connected to cellphones. It works like this: users call the service and then hold their cellphones up to the song on the car radio, and the system gets the song through the cellphone, comparing that audio with its database to find out the title of the tune.
This company has a database with the fingerprints of 1.6 million tunes. “It matches the incoming fingerprint with its database and within 30 seconds sends a text message back to the phone identifying the song.”
Philip Inghelbrecht, founder of the company and director of its content, said that if a CD is commercially available, they get it, so they have most popular songs that have been recorded.
This technology is useful for ordinary people who love music. According to Vance Ikezoye, chief executive of Audible Magic (this company presents its own patented audio fingerprinting technology), a company in Los Gatos, California, manufactures of electronic devices may present audio fingerprinting to people who want to arrange their music collections in their computers.
He said “It’s hard to manage music if you don’t have the correct information for every song in your collection.”
This article also explained one method for creating the unique code of the fingerprint system (companies have used different techniques to make this, as the article said). The method described was that used by Philips. According to Dr. Kalker, he and his colleagues made the code by breaking each tune into 10-millisecond pieces. Then, “they calculated the differences in the loudness of adjacent frequencies in the snippet and how those differences changed over time – they repeated the process every 10 milliseconds to extract code over the entire length of the song.”
Because of the precision of this database, Dr. Kalker said that the entire process, once the database is up and running, needs just about three seconds, so even if people listen to a bad AM or FM radio, it doesn’t bother the system to find name of the song.
Dr. Gooch mentioned the possibility of the fingerprinting system for use in car radios. He said, “when every car has its own digital audio player, people will want to know the name of the song they are listening to displayed on the dashboard.”
Some people might worry about the rights of music publishers when they use this technology. However, according to Mr. Ikezoye of Audible Magic, people don’t have to be concerned about violating the rights of music publishers.
He said that even though Audible Magic keeps a huge database of popular copyrighted music, these fingerprints are not the actual songs, just ‘summaries of factual measurements describing the sound.’
He also mentioned that the original sound cannot be rebuilt from the fingerprint, therefore saving and giving out fingerprints does not violate the copyright.
The article was concluded with a citation by Cary Sherman, president and general counsel of the Recording Industry Association of America: “Radios often don’t bother to tell listeners what they have played.” – the article said that if the technology can reach into consumer applications, radio listeners may solve the problem Sherman mentioned.
I think this Audio fingerprinting is notable, and I admire people who developed this technology. However, if this technology costs a lot in consumer applications, it might not be as successful as they think. Unfortunately, the article does not mention any projected costs or who will pay for the service.