Next  |  Prev  |  Up  |  Top  |  Index  |  JOS Index  |  JOS Pubs  |  JOS Home  |  Search

Experiences with the Samson Box

While the Music V software synthesis approach was somewhat general and powerful--a unit generator could do anything permitted by the underlying programming language--computational costs on a general-purpose computer were dauntingly high. It was common for composers to spend hundreds of seconds of computer time for each second of sound produced. Student composers were forced to work between 3 AM and 6 AM to finish their pieces. Pressure mounted to move the primitive sound-generating algorithms into special-purpose hardware.

In October 1977, CCRMA took delivery of the Systems Concepts Digital Synthesizer [5], affectionately known as the ``Samson Box,'' named after its designer Peter Samson. The Samson Box resembled a green refrigerator in the machine room at the Stanford Artificial Intelligence Laboratory, and it cost on the order of $100,000. In its hardware architecture, it provided 256 generators (waveform oscillators with several modes and controls, complete with amplitude and frequency envelope support), and 128 modifiers (each of which could be a second-order filter, random-number generator, or amplitude-modulator, among other functions). Up to 64 Kwords of delay memory with 32 access ports could be used to construct large wavetables and delay lines. A modifier could be combined with a delay port to construct a high-order comb filter or Schroeder allpass filter--fundamental building blocks of digital reverberators. Finally, four digital-to-analog converters came with the Box to supply four-channel sound output. These analog lines were fed to a 16-by-32 audio switch that routed sound to various listening stations around the lab.

The Samson Box was an elegant implementation of nearly all known, desirable, unit-generators in hardware form, and sound synthesis was sped up by three orders of magnitude in many cases. Additive, subtractive, and nonlinear FM synthesis and waveshaping were well supported. Much music was produced by many composers on the Samson Box over more than a decade. It was a clear success.

The Samson Box, however, was not a panacea. There were sizable costs in moving from a general software synthesis environment to a constrained, special-purpose hardware synthesizer. Tens of man-years of effort went into software support. A large instrument library was written to manage the patching of hardware unit generators into instruments. Instead of directly controlling the synthesizer, instrument procedures written in the SAIL programming language were executed to produce synthesizer commands that were saved in a ``command stream'' file. Debugging tools were developed for disassembling, editing, and reassembling the synthesizer command-stream data. Reading and manipulating the synthesizer command stream was difficult but unavoidable in serious debugging work. Software for managing the unique envelope hardware on the synthesizer was developed, requiring a lot of work. Filter support was complicated by the use of 20-bit fixed-point hardware with nonsaturating overflow and lack of rounding control. General wavetables were not supported in the oscillators. Overall, it simply took a lot of systems programming work to make everything work right.

Another type of cost was incurred in moving from the general-purpose computer to the Samson Box. Research into new synthesis techniques slowed to a trickle. While editing an Algol-like description of a Mus10 instrument was easy, reconfiguring a complicated patch of Samson Box modules was much more difficult, and a lot of expertise was required to design, develop, and debug new instruments on the Box. Many new techniques such as waveguide synthesis and the Chant vocal synthesis method did not map easily onto the Samson Box architecture. Bowed strings based on a physical model could not be given a physically correct vibrato mechanism due to the way delay memory usage was constrained. Simple feedback FM did not work because phase rather than frequency feedback is required. Most memorably, the simple interpolating delay line, called Zdelay in Mus10, was incredibly difficult to implement on the Box, and an enormous amount of time was expended trying to do it. While the Samson Box was a paragon of design elegance and hardware excellence, it did not provide the proper foundation for future growth of synthesis technology. It was more of a music instrument than a research tool.

Next  |  Prev  |  Up  |  Top  |  Index  |  JOS Index  |  JOS Pubs  |  JOS Home  |  Search

Download kna.pdf

``Viewpoints on the History of Digital Synthesis'', by Julius O. Smith III, Proceedings of the International Computer Music Conference (ICMC-91, Montreal), pp. 1-10,
Computer Music Association, October 1991.
Revised with Curtis Roads for publication in Cahiers de l'IRCAM, September 1992, Institut de Recherche et Coordination Acoustique / Musique.
Copyright © 2005-12-28 by Julius O. Smith III
Center for Computer Research in Music and Acoustics (CCRMA),   Stanford University
CCRMA  [Automatic-links disclaimer]