A. THE DOCUMENT
    -  pdf (8 MBytes)

B. THE DEFENSE
    -  html
    -  mpeg

C. SOUND EXAMPLES

1. Cluster-weighted spectral synthesis: A microphone was used to recod the audio data. Between 20 to 40 clusters and locally linear models were used to generate the models:
    example 1:  B-natural, detache bowing (original, predicted),
    example 2:  B-natural, sustained bowing  (original, predicted),
    example 3:  scale (original, predicted),

2. Cluster-weighted sampling.A dynamic pickup was used to record the audio data. 20 to 30 clusters were used to built the model:
    example 1:  B-natural, detache bowing (original, predicted),
    example 2:  B-natural, sustained bowing  (original, predicted),
    example 3:  scale (original, predicted),

D. VIDEO CLIPS

Take the money and run, Woody Allen, 1969.

Arm demo, summer 1998.

One-string violin, fall 1998.

Marching Cello, dynamic string response and mixed synthesis, summer 2000.

Stars and Strips Forever, Mary Farbood (Piano) and Bernd Schoner (Marching cello), summer 2000.
 

E. HARDWARE DOCUMENTATION

1. violin sensor board
    - schematic
    - PCB layout
    - PIC project,
    - PIC source code,