The most accepted church organ tone generation technology is digital sampling for a simple reason: a well-designed and voiced digitally-sampled church organ offers the most realistic pipe organ sound.  A comparison of digitally sampled stops and real pipe stops in the same room in a combination organ is the ultimate test of the quality of sampled voices.

While the earliest form of digital sampling was introduced in church organs over forty years ago, the technology has not been static.  Today’s digital organs include tens of thousands times the processing power and are as different from early digital organs as a modern laptop computer is from an early digital calculator.

Some organ suppliers attempt to differentiate their products by promoting their technology as new or different.  One company  promotes a technology based on physical modeling, claiming that modeling is new and sampling is older and therefore inferior.  This claim does not stand up to scrutiny.

First, physical modeling is not new as it was introduced in musical instruments decades ago by Yamaha (and since no longer offered by Yamaha).

In addition,  the same company that promotes physical modeling states: “Thanks to accurate audio reproduction and powerful reverberation effects …”  An organ’s acoustical enhancement is an important part of its overall sonic realism.  Digital reverberation has ceased to be state-of-the-art.  Today, the more advanced convolution technology, exactly replicating real acoustics spaces from performance halls to cathedrals, is state-of-the-art.

Irrespective of technology claims, it is ultimately tonal quality, detail and realism, that determine the value of any tone generation technology.  A high-quality digitally-sampled organ properly voiced is indistinguishable from a real pipe voice by even highly trained organists.  A modeled voice can, therefore, do no better.  But, in reality, physical modeling produces a less realistic result.

Sampling produces a precise duplicate of the original pipe sound and can, therefore, be used to exactly recreate the works of famous organ builders such as Schnitger, Silberman, Cavaillé-Coll, or Willis.  Because a modeled voice only approximates a pipe, the sound it produces is generic.  The choice is simple: modeling’s caricature of pipe organ sounds or sampling’s replication of real masterworks.

An internet comparison of digital sampling and physical modeling uses pictures to demonstrate the difference.  The one below titled “Physical modeling thing” shows an approximation of bricks generated through models.  The modeled bricks do not have the detail of the real ones included in the photograph; “Sampled thing”.  The author goes on to say: “So you might think… but we’re talking about musical instruments, not about a brickwall!!  Let me say that we’re talking about the same thing: the complexity of the natural forces involved in the structural formation of the reality.”



Viscount includes in its technological claims:  “with Physis, the parameters used to generate the sound are available so you can easily modify and mold the actual sound structure” and that the organs include: “……  a rich library of customizable settings and parameters”.  If the original pipe organ sounds are correctly reproduced and properly scaled and voiced by the organ builder, why would such changes be required?

The late world-renowned concert organist Carlo Curley wrote a fellow organist in November 2009 after playing an organ based on modeling: “Of course, before I heard and played the ‘Physis’, I was told by their salesman that this was the new system that will “change the digital organ world for the better forever”.  Also, “as it is so revolutionary, it will force all the other digital builders out of business, due to the extraordinary technology and pure quality of tone that is simply light-years ahead of their nearest competition”.

“I was to discover that for me, these claims were perfect examples of high-level hyperbole, meaning simply that their claims were obvious and intentional exaggeration and were also extravagant statements not intended to be taken literally.”

5 thoughts on “Tech Talk: Digital Sampling versus Physical Modeling

  1. Pingback: URL
  2. I’ve been tasked with deciding on a new organ model, and I’ve already talked with Rodgers-Johannus and Viscount; the Allen guy is due to come out tomorrow. I and our organist are both particularly interested in comparing the sound-generation technology of the four models, though, and I haven’t heard a really convincing reason for the sampling=good/physical modeling=bad argument.

    I think your analogy above doesn’t really work. An organ pipe produces vibrations, frequencies, waves, etc. which determine the quality of the resulting sound. All these things can be digitally analyzed, quantified, stored and reproduced if the software, the computer’s processing power and the output mechanisms are sufficiently sophisticates. If computers can successfully model and reproduce the behavior of a system as complex as a passenger aircraft, it seems completely counterintuitive to claim that they would some be less successful in reproducing the sound of something as simple as an organ pipe.

    Insisting that only sampling will do the right job seems like insisting that aerospace designers should go back to relying exclusively on wind tunnels.

    In videos on the attached playlist, by the way, you’ll see our original Allen. For its time, it had an excellent sound, and it was always a pleasure to play.

    Any insights on the above would be appreciated! Thanks.

    1. Thank you for raising this important question.

      As you point out, modeling has various applications in today’s high-tech world. It is often used when attempting to predict an unknown outcome that can be affected by many variables. An example is weather prediction. As a hurricane is building in the ocean, sophisticated computers use models (algorithms and inputted variables) in an attempt to determine where the hurricane will land and at what strength based. Unfortunately, as demonstrated in an event last month, these predictions can be inaccurate. Another example is the current difficulty that high-tech Boeing is having with the 737 Max. This is related to a modeling weakness. Boeing attempted through modeling to have the plane’s flight controls respond in a manner similar to that of the older 737s in order to avoid retraining pilots. The results have not been positive.

      Unlike a predictive outcome, with digital organs, the desired outcome is fully known. For example, should a digital organ builder want to produce an instrument based on a famous builder, such as Cavaillé-Coll, sampling enables the builder to take advantage of the genius/art of Cavaillé-Coll and re-create the sounds accurately, assuming a quality sampling system. Modeling, on the other hand, begins with generic pipes. Trying to create models that can “think” and act like Cavaillé-Coll is beyond current technical capabilities. This is a major reason sampling offers the best way to accurately reproduce pipe organ sound.

      It can be difficult to understand what kind of quality has been designed within integrated circuits/processors that are the heart of an organ’s tone generation. This Blog suggests that one method customers can use is by reviewing observable assemblies in the organ. It is reasonable to conclude that an organ builder who puts the highest quality in observable subsystems will do the same inside their tone generation system’s technology. Conversely, a builder who cuts corners with regards to quality in obvious ways will likely also do so in ways that are not transparent to the customer.

      Finally, the ultimate test of any digital organ’s tone generation technology is tonal results. Below is a link to a video that includes Allen Organ Company’s latest tone generation technology played along-side of a fine pipe organ in the same room and operated by the same console. Most who have heard this presentation cannot determine which sounds are produced by pipes and which are produced digitally. No matter what a salesman or technologist might claim, the results cannot be better than sounding like the real thing.

      Please do not hesitate to raise any further questions.

  3. Thank you for your lengthy reply. It will be very helpful to me in understanding both sides of the issue. You might want to consider putting a discussion of this technology into a video, as Viscount has for Physis. This would make it easier for the non-scientifically inclined (such as myself) to compare and understand the differences and similarities.

    You’re right, though — the proof of the proverbial pudding is in the sound.

    I’ll add here that I was always impressed by Allen’s organs, even in the pre-digital days when we had one in the choir practice room at the seminary.

    With cordial best wishes,

    Fr. Cekada

Leave a Reply

Your email address will not be published. Required fields are marked *