Isaak Vigdorchik (1929-1987)
Isaak Vigdorchik was a luthier, a performer, and a scientist who had emigrated to the United States from the Soviet Union after graduate school. In 1986, my cello teacher, James Hoffman, aware of my work in audio processing, suggested that I meet Isaak. That meeting planted a seed that has been germinating for four decades.
The Luthier and Scientist
In 1982, Isaak’s “The Acoustical Systems of Violins of Stradivarius and Other Cremona Makers” was published.
Isaak’s book was the culmination of thirty years of work as a performer and a luthier. In his career as a luthier he made over 100 instruments, mainly violins.
In Moscow, as a young student, Isaak had the opportunity to study under the well-known maker Stepan Dobrov where he learned how to determine the pitch of an instrument’s plate by tapping with his knuckles. He refined his technique while working at the laboratory of Denis Yarovoy where he was given the opportunity to examine rare instruments made by the Brescian masters, including some made by the Amati family.
From Isaak’s book
It wasn’t until 1977, when Isaak emigrated to the US, that he had the opportunity to study instruments made by Stradivarius who, as an apprentice of Amati, refined the master’s techniques and created the most prized instruments in existence.
Isaak’s keen tapping method and his pitch perfect hearing led him to hypothesize that the luthiers who built these unique instruments were able to shave the plates such that they would ring out/resonate at the 12 notes of the western musical scale. Page 99.
Top plate of Stradivari violin showing “sound strips”
Certainly an interesting idea. Could it be true???
We know now – thanks to laser spectrometry and other measurement and visualization technologies unavailalbe to him – that the modal patterns of a sound board have far more complex shapes than the strips Isaak suggests, but the idea of tuned chromatic resonances is no less valid. True or not, it got me thinking.
Recording studio reverbs were, at first, acoustic, from large concert halls to small echo chambers. When acoustic spaces were unavailable, unaffordable, or otherwise inconvenient, mechanically resonant devices were made to do the job – springs and plates. Digital simulations of spaces were an obvious idea, difficult to implement until the silicon could keep up with the calculations. When it finally happened, digital reverb designs were largely motivated by what was known about room acoustics. Rooms resonate. Plates resonate. Wait, so do musical instruments. Can musical instruments reverberate?
So, How Do I Fit In?
Call it happenstance, kismet, fate or simply dumb luck, but, in the early ‘80s, I was on a winding path to meet Isaak…
The SP2016
By 1982, Eventide was shipping our SP2016, a rack-mount studio device that was the world’s first general purpose audio processor. It was an ambitious design started in 1978, back in the early days of semiconductors, when ‘processors’ were built with ‘discrete logic’. The early CPUs and computers could not come close to doing any audio processing in real time. DSP chips were at least five years away, and the-computer-as-your-DAW was still twenty years in the future.
The SP2016 was designed to process audio as fast as possible at the time. It was a multi-chip array processor (Harvard architecture) constructed by combining the fastest discrete ICs available. The key components included a single chip 16X16 multiplier/accumulator designed for radar applications and a ‘custom’ 24 bit, fixed-point, ALU (Arithmetic Logic Unit) built by combining six ‘bit slice’ 2901s (4 bits each). Two banks of RAM, one static (for speed) and one dynamic (for size) completed the design of the array processor.
The unit could perform 128 computations per sample period of 20 uSec – powerful enough to do all kinds of things like vocoding, multitap delays, granular effects, band delays and, of course, algorithmic reverbs. New effects could be added by plugging in ROMs (Read Only Memory) into ZIF (Zero Insertion Force) sockets.
Eventide developed a programming language for developing algorithms; an early precursor to VSIG called SPUD (Signal Processor User Development). One of its developers, Steve Hoge, left Eventide and started First Order Effects and made some money developing new algorithms and selling plug- in chips with FX like “Inverse Reverb”, “Sync’d Repeats” and “Moving Reverb”.
With the SP2016 out in the world and the next step – the DSP chip – clearly just over the horizon, we found ourselves in a bit of a ‘holding pattern’. There was plenty of talk about a new breed of IC designed for audio processing that would replace the dozens of chips of my array processor. That chip, Texas Instruments TMS32010, was announced in 1983 and became the foundation of the H3000 Harmonizer – released in 1986.
In my spare time, I was trying to code a 1024 point Fast Fourier Transform to run on the SP2016 (using SPUD) but abandoned the task when I realized that, even if the transform ran, the processor would be out of cycles and couldn’t do anything useful to the audio. But I had made a good start at writing the micro-code and, as always, Richard’s most recent project triggered yet another idea…
The World’s First Computer-based Test Instrument – THS224
A year or so earlier, in 1980, Eventide’s founder and instrument maven, Richard Factor, was the first to demonstrate that a computer can be used as a piece of test equipment. He had designed an add-in card for one of the first personal computers, the Commodore PET, turning the computer into a third-octave real-time audio spectrum analyzer.
For the history of the THS224
Following (once again) Richard’s lead, it occurred to me that I could design (and sell) an add-in card for my computer, a Hewlett Packard HP9835. I knew that I could base the design on a simplified version of the SP2016’s array processor.
Back in the mid 70s, before personal computers became available, engineering companies, like Eventide, used desktop computers made by Hewlett Packard. We enjoyed a nice business selling add-in memory for the HP. It wasn’t something we had planned, but I needed the extra memory to run ray tracing software to study reverb. When we saw the price that HP was charging, we thought ‘easy money’. Simple design, great margin.
The FF523 – A Fast Fourier Transform Computer Add-in Card
Given our experience selling add-on memory and given the fact that the HP was in the hands of engineers and scientists worldwide, I thought that an add-in processor that supercharged the execution of FFTs would be useful. The FFT523 was a success – a processing card that enabled the computer to execute a 1024 point FFT at audio rate and beyond. It was used by engineers, physicists, and scientists working in fields from medical imaging to particle physics.
FFT523 – Add-in array processor for HP computer
For audio testing, its frequency resolution and precision were superior to anything possible in the analog world. In fact, Bob Berkovitz, who went on to found Sensimetrics, used it to create some of the earliest real-time audio spectrograms.
Meeting Isaak
Starting around 1982, I began taking cello lessons with James Hoffman, a brilliant teacher and performer, who lived near my home in Brooklyn. Mr. Hoffman, even though not a ‘gear’ person, had heard of the Harmonizer®. From time to time, he’d ask me what I was working on. In 1985/6, with an RTA application running on my FFT523-equipped HP computer, and likely underprepared for my lesson, I stalled and distracted him with a demo. Mr. Hoffman seemed intrigued and suggested that I meet his friend Isaak.
Perhaps he had read the closing remark of Isaak’s book?
While I’m certainly no physicist, I could provide the tool that Isaak would need to perform the kind of testing necessary to test his hypothesis. Analog third octave audio analyzers fell far short of the degree of precision required to test the theory propounded by Isaak. When I demonstrated the HP-based analyzer to Isaak, my recollection is, he smiled. And then he asked if I would help. I said yes of course! Cool project, great musician, and a great person. And the chance to hold a Stradivari? Doesn’t get any better than that.
Notes in the Wood
Isaak and I met a handful of times to discuss how we might work together to test his ‘conjecture’ and Isaak began to reach out to colleagues who could arrange for us to borrow rare violins. My task was to come up with a way to record the sound of his tapping directly to the computer and develop the analysis software. I confess I was somewhat skeptical of his theory, and I let him know that. Isaak would simply say, in his heavy Russian accent, “No no Anthony. The notes are in the wood.”
Notes in the Room
Sadly, Isaak passed away in 1987, shortly after we met, and so we were never able to even begin working together. But Isaak’s words echoed in my alleged brain and those words ‘notes-in-the-wood’ reverberated, morphing into “notes-in-the-room.”
Conceptually simple! Use a modal model to design a reverb that would change the sound of notes in the room by emphasizing the level and decay of the frequencies of the twelve-note scale. However, such a design would involve being able to control the ‘modes’ of the room; lots of them. Lots.
In 1985, I couldn’t imagine when, or even if, the necessary computational power would exist. Theoretically the notion of controlling the energy of notes by using a modal model of a room can seem straightforward; practically it remained impossible for decades; four in fact!
Here’s a peek at what we ultimately created.
