DIASS

a Digital Instrument for Additive Sound Synthesis
consists of a M4C type instrument and a score editor both written in C.

Chronology:

DIASS was designed by Sever Tipei and Christopher Kriese and the initial code was written by Christopher Kriese in 1991-92. Under the supervision of Sever Tipei, during 1994-96 David Ralley expanded the capabilities of the instrument and, together with Cheryl Herndon, added the loudness routines which allow the user to specify the perceived loudness of a sound event. A team including Dave Blumenthal, Mario Lauria, Thomas Lawrence, and Scott Pakin modified M4C and wrote a parallel version using MPI, a message passing library. Their work was done as a class project for Sever Tipei's seminar Musical Applications On Supercomputers offered in Spring 1995. During spring 1996 Dave Blumenthal and Max Levchin developed M4CAVE an OpenGL program which creates images in a 3D virtual environment (CAVE), from the same "score" that produces the sounds. Arun Chandra worked on improving the code during 1996-1997. Presently Mike Piacenza, Bill Whitehouse, and a group of Musical Applications On Supercomputers seminar are working on restructuring the code and at translating it into C++. Prof. Sang Lyul Min from Seoul National University has contributed substantially to the optimization of the parallel code (DIASS_M4C).

Hans Kaper, Senior Mathematician at Argonne National Laboratory, has been contributing to the restructuring of the program and has been supervising the project together with Sever Tipei since 1994.

General statement:

DIASS is intended to be an experimental composer's tool and, when designing it, the user's ability to control the details was given priority over cost. This translates into very large amounts of input data and intensive computations. A "Rolls Royce Bulldozer" , DIASS is intended to run on high performance computers. It was first implemented on the NCSA's CRAY Y-MP and presently runs on the IBM Scalable POWERparallel System (SP) of the Argonne National Laboratory Mathematics and Computer Science Division

Description of the instrument:

• DIASS can handle an arbitrary number of consecutive and/or simultaneous sounds of up to 65 partials each. Every partial can be independently controlled through static parameters (which do not vary during its duration) and through dynamic parameters (which may vary during the life of the partial).
• There are 12 static and 13 dynamic parameters. The static parameters include:
  • start time
  • duration
  • reverberation characteristics such as
    • hall size
    • duration of the reverberation.
  • phase, etc.
  The dymanic parameters include:
  • frequency
  • amplitude
  • magnitude and rate of:
    • amplitude modulation
    • frequency modulation
    • transients of amplitude
    • transients of frequency
  • panning
  • mix between direct and reverberated sound, etc.
• A number of macros apply certain operations to all partials making up a sound. They include:
  • glissando
  • tuning of a sound's partials according to desired ratios
  • dynamic tuning and detuning of partials during the duration of a sound
  • scaling of each partial's amplitude as a percentage of the fundamental's amplitude
  • enhancing or dampening of certain frequency bands similar to the effect the body of an acoustic instrument has on the resulting sound
  • perceived loudness of a sound, etc.
• A unique feature of DIASS is the scaling of amplitudes according to the desired perceived loudness of the resulting sound event. The Fletcher-Munson curves are used to determine the amplitude necessary to produce a certain perceived loudness in phons. They are further adjusted taking into consideration the critical bands whithin which they might fall (see references below).
• The anticlip option insures (before the samples are computed) that there will be no "clipping" in the sound file. If one or more samples exceed the maximum amplitude allowed, the entire piece will be scaled accordingly. Except that, in stead of lowering the amplitude across the board, each different combination of simultaneous sounds is evaluated by the loudness routines and each partial is re-scaled according to the Fletcher-Munson curves and to their position within critical bands. Practically, this results in the fact that works with a wide dynamic range can be produced in one run of the program without resorting to "post-production" digital or to analog mixing.
• There are two versions of the score editor, diassin and fast, and two versions of the M4C instrument, sequential and parallel.
  • diassin or the slow version of the editor allows the user to experiment with creating new sounds through a menu or through a Graphic User Interface (GUI) presently under construction. The process is slow due to the large number of options available to the user.
  • The fast version of the editor is recommended in production mode or when using the output of a computer-assisted composition program. A script provides the answers to the questions posed by the menu of the slow version.
  • As the name indicates, the sequential version of the M4C instrument requires only one processor.
  • The parallel version of the instrument uses MPI a message passing library to divide the computations between a number of nodes. All nodes except two compute one sound at a time each until the piece is finished. One of the remaining nodes has the task of distributing the work load and the last one mixes together the computed sounds as they arrive.

Works produced with DIASS:

• Sever Tipei
  1. AGA MATTER for piano and computer generated tape, 1992.
  2. RICE MATTERS for computer generated tape, 1993.
  3. CURSES for solo male voice, backup group and computer generated tape (some of the sounds), 1996.
  4. ANL-folds manifold compositions for computer generated tape, 1996 -
  5. Sonic Residues, for computer generated tape; 50 variants were performed on December 21, 1997 at the Linden Gallery, in Melbourne, Australia.
  6. BERLIN-folds #1, #2, for computer-generated tape (1998)
  7. A.A.-folds, for computer generated tape, an installation at Int'l Computer Music Conference 1998, Ann Arbor, Michigan.

User's manual:

In the future you will be able to click here to read the user's manual presently under construction. Until then, you can see a list of Controls and Inputs and DIASS I-Card Description.

References:

1. Christopher Kriese and Sever Tipei - A Compositional Approach to Additive Synthesis on Supercomputers, Proc. of the 1992 Int'l Computer Music Conf. (San Jose, California; September 1992), International Computer Music Association, San Francisco, CA, 1992, pp. 394-395.
2. Hans Kaper, David Ralley, Juan Restrepo, and Sever Tipei - Additive Synthesis with DIASS_M4C on Argonne National Laboratory's IBM POWERparallel System (SP), Proc. of the 1995 Int'l Computer Music Conference (Banff, Canada; September 1995) International Computer Music Association, San Francisco, CA, 1995, pp. 351-352.
3. Hans Kaper, David Ralley, and Sever Tipei - Perceived Equal Loudness of Complex Tones: a Software Implementation for Computer Music Composition , Proc. 1996 Int'l Conf. in Music Perception and Cognition (Montreal, Canada; August 1996), pp.127-132.

4. DIASS was also presented at the SuperComputing'95 (San Diego, California; December 1995), SuperComputing'96 (Pittsburgh, Pennsylvania; November 1996), SuperComputing'97 (San Jose, California, November 1997) conferences as part of the Argonne National Laboratory exhibits as well as at the ICAD (Palo Alto, California; November 1997) conference.

DIASS was developed with funds provided by the Research Board of the University of Illinois at Urbana-Champaign and by the Mathematics and Computer Science Division of the Argonne National Laboratory
Both MCS/ANL and NCSA provided time on their supercomputers.