Kinesthetic Calisthenics

Not being a keyboard player, I've always been interested in alternate controllers. Such controllers, however, often suffer from inherent deficiencies
Publish date:
Updated on

Not being a keyboard player, I've always been interested in alternate controllers. Such controllers, however, often suffer from inherent deficiencies that limit their usefulness as expressive musical instruments. For example, many players of string-based MIDI controllers complain about a lack of expressive capability caused by a sense of disconnection between their gestures and the music they are trying to play. This is especially apparent when the controller makes no sound of its own.

To address this problem, Charles Nichols (, assistant professor of composition and music technology at the University of Montana, is developing a new type of bowed-string controller, called the vBow, as part of his doctoral work at Stanford University's Center for Computer Research in Music and Acoustics (CCRMA). His original idea was to design a controller that mapped a violinist's gestures to various synthesis parameters. However, he discovered that isn't enough to provide a complete sense of connection with the music.

After studying experiments in kinesthetic, or haptic, feedback conducted by Sile O'Modhrain and Chris Chafe at CCRMA, Nichols realized that such physical feedback improves a controller's playability and the performer's response to the system. As a result, the vBow not only controls sound parameters according to various bowing gestures, but it gives kinesthetic cues to the performer, similar to what they would feel playing a real violin.

In the current incarnation of the vBow, several pieces of custom-milled acrylic are assembled with servomotors, cables, and capstans into a mechanism mounted on a violin-shaped base, and a fiberglass stick resembling a violin bow passes through a hole in one of the pieces (see Fig. 1). Encoders attached to the servomotors sense the bow's movement in four degrees of freedom: lateral (across the “strings”), rotational (which “strings” are being played), vertical (degree of pressure on the “strings”), and longitudinal (along the length of the “strings”). It's important to understand that the vBow simulates the sound and feel of bowing strings; there are no physical strings involved.

The encoders send their data to a computer running software that applies the information to various parameters of a bowed-string physical model. In addition, the computer sends control voltages back to the servomotors, which impede the bow just enough to simulate the haptic feedback of a real violin; specifically, friction (lateral and longitudinal), vibration (lateral), string-crossing detents (rotational), and string elasticity (vertical).

The physical-modeling synthesis software was written by Stefania Serafin using Synthesis ToolKit, a collection of C++ objects developed at CCRMA for generating computer-synthesized sound. The control software, which provides the haptic feedback, was also written in C++ using libraries supplied with the ServoToGo servomotor-control and data-acquisition card used by the computer to communicate with the vBow.

The latency imposed by Nichols' current computer, a 600 MHz Pentium III running Windows 98, is quite noticeable. As a result, he intends to port the whole thing to a faster processor running RTLinux to reduce that latency and ensure nearly real-time performance. In addition, the system is currently limited to the four open “strings” of a violin; future versions will include vStrings that will track finger positions to derive pitches. Clearly, the vBow is a bold step toward a new type of bowed-string instrument controller that will be far more expressive than anything available today.