University of Rochester
[NEWS AND FACTS BANNER]
NEWS AND FACTS

Skip Navigation Bar
July 28,
2003

Contents

Previous article

Next article

In Brief

Calendar

Classifieds

Jobs

Currents home

Mail


Phone BookContact the UniversitySearch/IndexNews and Facts
 
Currents--University of Rochester newspaper

Software delivers 'mood music' on demand

Ogihara
Ogihara

It's a rainy afternoon, cappuccino's in hand, and all that's needed now is some mellow jazz to set the mood. Well, Mitsu Ogihara is developing a new software program that may be able to oblige in the search for just the right tune.

The professor of computer science has applied for a patent on software that not only can tell whether a piece of music is jazz or classical or rock but also can detect a song's emotional tone. Ogihara hopes that the program will one day allow people to surf the radio or Internet for music across genres that matches the listener's specific mood.

"We're looking to close that last gap between computers and music," says Ogihara. "Computers play music, download it, record it, manipulate it, but they're terrible at having a feel for it."

Developed with the assistance of graduate students Tao Li and Stephen Li, Ogihara's software works differently from most programs previously designed to detect musical genre. Such programs typically analyze drum beats, melodies, or complexity, but Ogihara's program scans about 30 seconds of a piece of music and looks strictly at statistics gathered from the sound's waveform. The program analyzes how often similar spikes happen, the relationships between certain parts of the waveform, and other characteristics to build a statistical profile.

As it turns out, the emotional content of music correlates very closely with these relationships, so Ogihara's software, while still in the early stages of development, is able to correctly identify emotional evocation 64 percent of the time and categorize genres with an unprecedented 78 percent accuracy. By comparison, a recent study of humans' ability to sort music into genres revealed only 71 percent accuracy.

Although initially successful, Ogihara says he'd like to improve the software's precision rates. "We'd like to push the precision to 90 percent before we think about bringing this to industry."

The software is generating interest for a wide range of applications. The Library of Congress hopes Ogihara's software may help its members more efficiently categorize music by genre, something that takes a tremendous amount of time and energy when compared to categorizing a book, which generally has all its relevant cataloging information on the book's sleeve.

For the average person, the software could help organize personal MP3 collections and sort songs into appropriate genres.



Maintained by University Public Relations
Please send your comments and suggestions to:
Public Relations.

 
SEARCH:     Directory | Index | Contact | Calendar | News | Giving
                     ©Copyright 1999 — 2004 University of Rochester