Media Releases >
Media Releases Archive
> New technology offers promise to speech therapist and reluctant singers
New technology offers promise to speech therapist and reluctant singers
Document Tools
June 2, 2003
David Gerhard heads a band called Lassiter Road, a trio of singers/musicians who fuse Celtic, folk, jazz and rock.
However, that’s just for fun.
Gerhard does sing for his supper – using computers. One of SFU’s latest crop of PhD graduates, Gerhard designed a way to measure music to quantify differences between singing, reading poetry, speaking and even rapping. For example, with vibrato (think Joan Baez), where the voice wobbles slightly, Gerhard numerically rates to what degree this feature is present. "It’s not something a lot of people have done before," Gerhard says from the University of Regina, where he is now an assistant professor of computer science. "No one has tried to quantify these phenomena into measurable differences."
Gerhard, who received many scholarships, fellowships and awards while at SFU, says his research has numerous practical applications. It is one small, but significant, step in further laying the groundwork for the time when voice-recognition programs in computers will detect subtle nuances in a person’s voice and react with an appropriate response.
More immediately, singing teachers, musical educators and speech therapists would benefit from systems like Gerhard’s that quantify sounds. It would also open new vistas for songwriters who produce electronic music on computers, allowing them to create new notes by blending speaking and singing.
Gerhard’s research also has implications for Query by Humming, a system of computer music recognition that is still more theory than reality. A person hums a few bars of music into a computer, which then provides the name of the song and links to sites where it is offered for sale on CD, or where it can be downloaded.
Ultimately, does this mean a tin ear can learn to sing like Luciano Pavarotti, or Celine Dion? Not likely, but at least Gerhard’s creations could determine how one’s voice differs from these famous singers. "That’s a first step to figuring out how to change your voice, so it could sound similar," Gerhard says.
So for all those paranoid about public singing (something else Gerhard is exploring) there is hope. Just don’t count on a Grammy.
-30-
Contact
Carol Thorbes, Media & PR, 604.291.3035; cthorbes@sfu.ca
However, that’s just for fun.
Gerhard does sing for his supper – using computers. One of SFU’s latest crop of PhD graduates, Gerhard designed a way to measure music to quantify differences between singing, reading poetry, speaking and even rapping. For example, with vibrato (think Joan Baez), where the voice wobbles slightly, Gerhard numerically rates to what degree this feature is present. "It’s not something a lot of people have done before," Gerhard says from the University of Regina, where he is now an assistant professor of computer science. "No one has tried to quantify these phenomena into measurable differences."
Gerhard, who received many scholarships, fellowships and awards while at SFU, says his research has numerous practical applications. It is one small, but significant, step in further laying the groundwork for the time when voice-recognition programs in computers will detect subtle nuances in a person’s voice and react with an appropriate response.
More immediately, singing teachers, musical educators and speech therapists would benefit from systems like Gerhard’s that quantify sounds. It would also open new vistas for songwriters who produce electronic music on computers, allowing them to create new notes by blending speaking and singing.
Gerhard’s research also has implications for Query by Humming, a system of computer music recognition that is still more theory than reality. A person hums a few bars of music into a computer, which then provides the name of the song and links to sites where it is offered for sale on CD, or where it can be downloaded.
Ultimately, does this mean a tin ear can learn to sing like Luciano Pavarotti, or Celine Dion? Not likely, but at least Gerhard’s creations could determine how one’s voice differs from these famous singers. "That’s a first step to figuring out how to change your voice, so it could sound similar," Gerhard says.
So for all those paranoid about public singing (something else Gerhard is exploring) there is hope. Just don’t count on a Grammy.
-30-
Contact
Carol Thorbes, Media & PR, 604.291.3035; cthorbes@sfu.ca