... essentially a set of math equations, which describe how vibrations would occur within a physical object.
Traditional approaches in sound synthesis recreate the spectrum of musical instruments by additive synthesis or frequency modulatio (FM), or by use of wavetables and samples. Unlike theses approaches:
Physical Modelling does not primarily focus on the produced sound itself, but on the behavior of a model (of an acoustic instrument) producing sound.
The two main approaches to instrumental sound synthesis today are sampling of acoustical instruments and physical modelling. While sampling offers excellent sound quality, but provides less control of dynamic variability, physical models are expressive and dynamic, but usually quite complex to understand and program, especially when a great range of dynamic variability is attempted to be controled.
I am interrested here mainly in PM to be played in realtime with a wind-midi-controler (there is plenty of other PM around).
PM is often seen as mainly targeting a realism in sound. From my perspective as a musician the realism of synthetic instruments is not oriented at sounding like an acoustic instrument, but at behaving like one during performance. This does not mean that it feels like a cello or sax, but the sound-synthesis can be controled via a controler (e.g. breath controler) so that all the nuances of the voice can be accessed in a natural way.
PM is very computing intesive, because the equations that describe a sound on the basis of a model has to be calculated in realtime. Therefore strategies are used to achieve realtime performance. e.g. socalled waveguides are basic instrument-behaviors (simplification of travelling waves as delay line structures) that are preproduced for each voice to be used then in realtime as a lookuptable. Attempts to overcome restrictions of waveguide-model, see e.g. multidimensional transfer function.
Another approach Modal Synthesis is used by the Tassman softsynth annd is basically a collection of parallel high-resonance bandpass filters, excited by an input impulse. The filters model the resonances in objects such as strings, pipes, bars, plates, and membranes. Each filter has three parameters: frequency, resonance (decay time), and output level. It is especially good for simulating struck sounds, like rods, bars, and plates.
See also Physical Modelling Particles, which is said to be cheap in computation, has meaningful parameters, but is not exact and not complete (see e.g.: PhISEM). Other approaches
try to establish a model by recording the playing parameters together with
the audio-output of an acoustical instrument and work with neural networks,
or cluster-weighted modelling (e.g. Schoner,
e.a. MIT, Data Driven Modelling and Synthesis..).
Intro by Scott Lehman: Physical Modeling Synthesis
Intro by Dennis Normer: Physical Modeling
Intro to Waveguide: Wikipedia
VL1-programming intro/tips: Manny Fernandez: VL PROGRAMMING GUIDE AND TUTORIAL
Julius Orion Smith III (founder of waveguide) : Digital Waveguide Modeling of Musical Instruments
PHYSICAL Modelling ccrma-www.stanford.edu/CCRMA (obsolete link)
Albino, Absynt, Gladiator
Applied Acoustics : Tassman, Lounge Lizard, Ultra Analog, String Studio, Strum Acoustic
Arturia Brass
Electric Grand EG70
Harm Visser toolkit for Reaktor and OASYS PCI
MilesTone and Sax-o-Phunk (Guido)
Modalys (for Power Macintosh) and its version with a graphical interface, Modalys-ER,
is used for creating virtual instruments based on simple physical objects such
as strings, metal plates, tubes, membranes, plectra, reeds and hammers. The
user “constructs” an instrument and then defines how it will be
played. Modalys-ER allows the user to dialog with Modalys, the synthesis engine,
via an intuitive graphical interface.
PM www.acustica.org www.audionica.com/Audionicaen/synthes.htm
Modartt Pianoteq (I use this software: impressive)
pmpd PM for pd
Reaktor PM
Revitar
STK (Scavone) Tutorial
Tao, Mark Pearson
Wallander Instruments WIVI etc
Yamaha S-YXG100 plus VL and S-YXG1000 plus PolyVL (the latter released in Japan only, 1999). These were basically software-only equivalents to the hardware (and hardware-assisted software) MIDI synth capabilities of the DS-XG cards / YMF chipsets. The PolyVL had eight voice polyphony for the physical modeling, whereas the VL and all of the hardware Yamaha VL synths only had one voice, or two for the original VL-1. Like the DS-XG .VxD drivers required for VL support of the DX-XG chipsets, these would work only on pre-NT kernel versions of Windows (9# and ME), and not on NT, 2000, XP, etc. Designed to take advantage of the performance of Intel’ s Pentium® III processor. There is a way to use s-yxg100 in xp, thru wdm s-yxg50 drivers, or using paralells and driver. (WinHex for fixing the zero-byte-at-end-of-file omitting bug, and set the date before 2001 on your Win98 machines). According to documents from yamaha, s-yxg100pvl should allow the editing of vl voices and assignment of parameters with one of the plugins that is installed (VA-XG ?). The system takes the exiting S-YXG50 engine and adds 256 preset VL voices conforming to the VL/XG spec of the VL70m. It also combines the Voice mode programmability and parameters of the VL70m. Yamaha quietly discontinued these years ago (excerpts from net, somewhere. I find this interresting, since there is a software-implementation at least of the VL70, which makes it more likely that some day there will be perhaps an even improved software-VL1).
Korg Oasys (PCI)
Nord Modular G2 (also PM, wavegide model, tutorial)
Roland's COSM technology in the VG-8
Yamaha VL7, VL1, VL1-m : (vl1/7 is a keyboard, vl1-m a rackmount, vl7 is a mono-voice vl1/-m dual-voice) all discont.
Yamaha VL70-m: reduced, mono-voice version of VL1
UNDER CONSTRUCTION