Not all of us are born with vocal chords of Andrea Bocelli or Sarah Brightman. Of course, great opera singers need more than natural ability. Many years of vocal training are required to perfect the operatic singing voice. However, some scientists are trying to make it possible for even the most tone deaf among us to sound like one of the Three Tenors or Sarah Brightman.
On Wednesday and Saturday nights at the Stained Glass Pub in Silver Spring, Maryland regular folks have a few drinks and try to sing some of their favorite songs. Some people are pretty good at it. Others need a bit more work.
An admitted lack of natural ability inspired Mark Smith, Head of Purdue University's School of Electric and Computer Engineering to see if he could eliminate the need for voice lessons. "Before you begin any work, there's some little motivating factor that's hidden in the back of your mind. For me, as early as 1985, I thought it would be great if there were a computer that could make me a great singer. Personally I would love to have something like this," he says.
Pop music in the 1980's relied heavily on electronic synthesizers to create music and digitally mimic acoustic instruments. The instruments' sounds were all copied from recordings and later manipulated by the synthesizer. A little computing power can affect the pitch, the duration of a note and in some cases, its quality.
Those are some of the features that Professor Smith wants to manipulate in human voices using the same techniques as a synthesizer. Believe it or not, he says that changing a singer's pitch is not a problem.
High-end recording studio software has been able to do this for a while now. Once upon a time, pop stars had to record another take if they were off key. Now, with the touch of a button, the pitch of artist's voice can be corrected, saving time and money, but also raising some ethical questions.
In the less artistic world of Karaoke, companies like Taito of Japan are studying the possibility of fixing singers' voices in real-time, using MIDI sequencing. MIDI, which stands for Musical Instrument Digital Interface, allows the computer to recognize the notes of a digital piece of music. Combining that technology with Professor Smith's work could allow a computer to change the singer's voice to match the notes people are supposed to hear.
The Stained Glass Pub may never sound the same again, but Mark Smith isn't stopping there. "The ability to change the pitch has been around for many, many years. This is not new. What is the unknown is the ability to address quality. We can't even define quality. We don't have numbers that can address quality," he says. "You can't say I want the singer to be number A-54320. We just have these kinds of subjective impressions of what sounds good."
That's not to say Professor Smith and his students haven't tried. As you can hear, not only has the pitch of the singer improved, vibrato has been added. "One of the things that's very interesting and we're not there yet at all is, 'How do I get an opera singer to sound like a classical singer or a Broadway singer?' These are questions that are fascinating to us and we're trying to understand that a little bit better," he says.
There could be commercial applications for digitally changing singing styles. A pop singer could release a country version of a successful pop album without stepping back into the studio.