Capturing the elusive mechanics of the violin
A sensor-studded violin bow offers a new way to understand how string players make music. The Hyperbow, designed by Diana Young, can describe and document the subtleties of a player’s technique. It could make synthesized violins sound more realistic, lead to new teaching tools, or preserve great musicians’ techniques.
“I think it would be exciting to create an archive of living artists, to preserve the individuality of their technique for future generations,” says Young, who at press time was defending her PhD in Media Arts and Sciences.
A violinist herself, Young knows there are different ways to achieve distinct sounds on the instrument. For example, a player can vary either the speed or pressure of the bow to control volume. Such techniques are universal, but each violinist uses them in a unique way. It isn’t easy to pin the differences down, much less to replicate them.
“The interaction between the bow and the string is too complicated to know exactly what [the musician] is doing by the audio alone,” Young says.
To capture data about violin technique, Young mounted off-the-shelf sensors on a carbon-fiber bow. Accelerometers and gyroscopes on the heel of the bow capture information relating to speed; two strain sensors mounted in the middle of the bow measure downward and lateral force; a custom-made electric-field sensor behind the violin’s bridge receives signals from the tip and heel of the bow to determine the distance between the bow and the bridge. As a violinist plays, the music is recorded, but not just with audio equipment.
The sensor data from the bow are transmitted wirelessly to the violin, where they are combined with data from the electric-field position sensor and sent to a computer through a cable. The data are fed into a program that imitates the sound of violin strings; the program not only plays back the correct notes but also duplicates the musical nuances of the player’s performance.
This is a big leap from existing synthesizers, says Julius Smith, professor of music and associate professor of electrical engineering at Stanford University. Current synthesizers, he says, may offer very convincing violin sounds derived from recordings, but they do not capture the musical variations introduced by human performers.
“Really good players are way beyond what we know how to write software for,” says Smith. “It’s like writing software to speak.”
The Hyperbow could lead to software that lets music producers modify recorded violin solos so that they mimic famous virtuosos’ styles. But that will require smoothing out one technical wrinkle. The bow’s sensors generate raw data that must be converted into usable measurements, like meters per second or newtons–a time-consuming, labor-intensive task.
The AI revolution is here. Will you lead or follow?
Join us at EmTech Digital 2019.