Introduction --

I always liked to design and make myself any kind of mechanisms and electrical and electronic equipment.  Starting at the age of nine I have been making radio receivers, transmitters, electric vehicles, hi-fi equipment, etc., and especially, musical instruments, including pipe organs.

When I discovered player and reproducing pianos some 20 years ago, I loved them at once.  I acquired knowledge of pneumatic technology and I restored many player and reproducing pianos (Duo-Art, Triphonola and Welte) and made various instruments and devices, such as:
 

I knew the existence of Mills Violano and Hupfeld Phonoliszt-Violina.  So, why not make a self-playing violin similar to those machines?  The wide availability of MIDI files with good violin performances would make the job easier.

Experimentation and technical information --

I decided to make a machine similar to Mills Violano, being much simpler than the Hupfeld instrument.  The use of electromagnets, instead of pneumatics, would make possible a more compact and simpler machine.

I soon understood that one of the critical matters of the job were the bow wheels.  I tried various materials (wood, rubber, plastic, metal), but the sound and its attack was poor.  The bow wheels could not emulate a real violin bow, that uses horse hair to provide many tiny rubbing points.  The original bow wheels, and the ones by Ralph Schultz, use many extremely thin and highly flexible plastic discs.  It would be difficult for me to make them, so I had them made by Ralph.

Another problem was to emulate the human finger fretting the strings.  An imperfect fretting would produce bad sound, overtones, harmonics.  I avoided to fret the strings from under them, like it happens in the Mills Violano, because this would have required important modifications in the violin.  So I put metal fingers fretting from above, but they did not work properly, because the vibrating strings were not sufficiently stopped by such metal fingers.  So I used soft plastic on them to act as a brake for string vibration.

The compass of the instrument is 48 notes, G3 to F#7 -- higher than the Violanos span in order to give a better performance of virtuoso pieces using very high treble notes.  The first three strings produce only seven notes each.  This is a big simplification but it does not cause important limitations, because no note is missing in the scale.  Only some two-string chords in some piece would be skipped, or could require editing of the MIDI file to make them play.

The distance of the fingers to give the correct musical scale was calculated by a computer program.  The distance between the fingers decreases as pitch increases, and is very small at the top of the scale, requiring high precision manufacturing.

The performance is made more realistic by a vibrato device: a strong solenoid shakes the violin tail piece, acting automatically and gradually after an initial delay, so to affect only the long duration notes.  The bow wheels are rotated at very low turning speeds of 100 to 300 rpm.

The self-playing violin is controlled by a MIDI decoder and can play, from MIDI files, any type of solo and ensemble music, with the accompaniment of a player piano (with MIDI interface) or a Yamaha Disklavier, or an electronic sound generator, or other MIDI instruments.

I am now studying a cello and a bass viol to play with the violin.  In the future I might also make a string quartet (two violins, viola, cello).

Thanks to --

I have to thank Robbie Rhodes and the MMD, by which I learned many things about mechanical music.  A special thanks to Ralph Schultz, an expert restorer of the Mills Violano, who built for me the bow wheels and gave a lot of technical information.  Thanks also to Silvestro, here in Rome, Italy, for the useful advice and the accurate making of some important mechanical parts, such as the bow wheel shafts.

Leonardo Forti
Rome, Italy
leonardo.forti@virgilio.it
31 May 2003 08:52:19 -0000

MP3 audio files:
  forti_czar.mp3 (371 kb, playing time 1:34, 32 kbit/s, 11025 Hz, Stereo)
  forti_paganini1.mp3 (373 kb,  playing time 1:34, 32 kbit/s, 11025 Hz, Stereo)