Sound and Music Conference

The first paper about Vivi was published in the Sound and Music Conference 2011. It is published under a Creative Commons license:

conference paper

slides for talk

Many of these examples may be found elsewhere on this site, but I present them here in the same order in which they are mentioned in the paper.

Audio 1: selected pieces

Black-box Testing

Graham Percival

lilypond source pdf mp3 mpeg

The Happy Farmer

R. Schumann (1810–1856)

lilypond source pdf mp3

Gavotte with piano

F.J. Gossec (1734–1829)

lilypond source pdf mp3

Audio 2: non-deterministic elements

The physical modeling includes random numbers in the calculation of bow friction. This means that the same physical parameters (as input) can produce very different output. In these examples, we recorded the physical actions from the first performance (with the control loop), then re-rendered the audio two more times. The differences are not extreme, but they are noticeable.

Original scale (with control)


Copy 1 scale (no control)


Copy 2 scale (no control)


Audio 3: Sound classification

Vivi asks you to classify short (approximately 0.5 seconds) bits of audio. These are created without any attacks, and are intended to test her "steady" sounds.

The sounds are classified as follows:

1. Much more force


almost silent!

2. More force


3. Ok sound


"ok" in comparison to the other sounds

4. Less force


5. Much less force


not a violin?

Audio 4: effects of training

Interactive training dramatically improves Vivi’s bowing control. This is particularly evident on the E string.

Example of training scales

Number of audio snippets trained

Approximate time spent

Audio examples


5 minutes (basic training only)

separate scale mp3 slurred scale mp3


15 minutes

separate scale mp3 slurred scale mp3


40 minutes

separate scale mp3 slurred scale mp3