The Garritan Corporation designs software tools in the form of sound libraries to aid in music composition and production. Founded by Gary and Marianne Garritan in the late '90s, it became one of the most successful sample library companies in the world. I joined as Director of Programming a couple of years after its inception. I remained with the Corporation for 10 years until my retirement.
I was Director of Programming, later doubling as Vice President, and then, finally, doubling as President of the Garritan Corporation for the final year before my retirement. I designed and programmed sample libraries for composers. A sample library is computer software consisting of musical instruments that have been recorded note-by-note (sampled) and programmed into a form (virtual instruments) that composers can control from within other host computer software or run as a standalone application. These virtual musical instruments include, but are not limited to, all standard acoustic categories like violins, violas, cellos, basses, harps, oboes, bassoons, clarinets, trumpets, trombones, tubas, timpani, cymbals, pianos, organs, drums, etc. Composers can choose, within the software, the desired instrument(s) and then compose music using MIDI data or notation to trigger these emulative counterparts, thus allowing them to hear the music as they compose it. I designed these typesof libraries relying upon my experience as a musician andcomposer to make useful decisions about their construction. These compositional tools included the Garritan Orchestral Strings, Garritan PersonalOrchestra, Garritan Jazz & Big Band, andGarritan Concert & Marching Band. They were given numerous awards throughout the music industry.
Go to Garritan Sample Libraries for information, demos, awards, and endorsements beyond what I show here. The Garritan Corporation is now owned by Makemusic (producers of the premiere music notation program Finale) of Minneapolis, MN.
An Unexpected Tangential Opportunity
To understand my Garritan association we must travel back to that far away mythical place called the late 1990s when a new music sampling technology from a company called Nemesis was introduced. The software technology was named Gigasampler. For years samples had been limited by static RAM memory size in hardware samplers. The total RAM size determined how much sample material could be loaded for use at a given time. Basically, that meant (in technical jargon) “not much” because RAM size for the time was on the order of 640 kilobytes and 1, 2, 4, or 8 megabytes. To clarify, "Mega" is the official scientific/mathematical term for "itsy-bitsy" and "kilo" is the term for "really itsy bitsy." For instance, a piano with 88 keys required far more memory to record the full decay of each note (which could exceed 40 seconds for lower notes,) and each note at different volume levels, than any sampler could make available at the time. So, tricks became necessary to work around these "I want to squeeze your pickup truck into my back pocket" limitations. One workaround was to record only a selection of notes at evenly spaced intervals across the range of the instrument and “stretch” transpose, up and down a few pitches, the actual recorded notes to act as stand-ins for the omitted notes. If the "stretch" was kept to a minimum that trick worked pretty well. Another was to record only the attack portion of the note plus a small part of the sustain or decay and then “loop” (endlessly repeat) this small part while (in the case of an instrument like a piano) applying an artificial envelope to simulate the rest of the full decay of the note. This one was a weaker choice but helped. Usually, a combination of these two (and other) techniques was used to get the maximum mileage out of the severely limited RAM available.
Into the fray came Gigasampler. Gigasampler changed everything by introducing a method of streaming audio data directly from a hard drive in real time and hard drives had far more space available than any configuration of RAM at the time. This effectively removed most of the practical limitations to sampled instrument design. For the first time all notes of a piano could be sampled at multiple levels with the full natural decay included. With and without sustain pedal even! This was a new paradigm.
Gary Garritan is a professional harpist. See photo on right where Gary is trying his level best to look very young. I was going to say he looked especially plucky in this photo but decided against it as my fading better judgement took over. Fortunately, that key decision saved me from having to back pedal later by a couple of half steps up or down. He introduced one of the first native libraries for the Gigasampler format – a concert harp called "Gigaharp." I was quick to purchase this new instrument and proceeded to do what I had always done with any commercial sample library: I modified the programming to do what I personally needed from the instrument to most easily facilitate my work. This had become my habit from the earliest sample collections and libraries I acquired for my Yamaha and, especially, Akai samplers. I had very specific requirements for the work I was doing with PBS and had become adept at making specific programming modifications or even rebuilding instruments from the ground up. Sometimes I would record my own samples, if nothing acceptable existed commercially. In the case of the Gigaharp I decided to send my programming modifications to Nemesis (the makers of Gigasampler and distributors of Gigaharp) and told them to feel free to use the changes since others might find them useful. Nemesis passed my changes along to Gary. Gary to a look and called me on the phone and threatened to sue me - no, no, no. We had a nice conversation about my programming modifications, which he seemed to find interesting.
A year passed. In early 2001, I got another phone call from Gary telling me that he had just recorded and edited a large group of top notch string section samples (violins, violas, cellos, and double basses of the highest quality) and wondered how I would approach programming a new string library, if given the chance? This began a couple of weeks of talks with me and other candidates for the job - an idea-fest between Gary and the potential programmers. My ideas were somewhat at odds with the other programmers concerning programming choices to best utilize the intended sampling platform's capabilities. These differences of opinion led to Gary visiting my studio where I showed him files I had prepared illustrating the sample platform behavior that had formed my opinions. In other words, I tried to demonstrate why I thought my opinions were the correct ones (and the others were wrong.) Gary recognized instantly that I was that rare type of individual who was always right . . . that is, I mean, apparently, Gary found my demonstration convincing because he decided to hire me to program The Garritan Orchestral Strings (GOS.) By this time Gigasampler had evolved into Gigastudio and GOS was introduced on that platform. Nemesis was later purchased by Tascam and then, eventually, the Nemesis assets were acquired from Tascam by Garritan. Small world.
To the right: The mixer page of Gigastudio, the software platform I used to program the Garritan Orchestral Strings.
Nemesis' Gigastudio 160 sample streaming software effectively removed many of the limitations for sampled instruments, not the least of which was total sample size. A tiny amount of data at the beginning of each sample was buffered while the remainder of the sample was streamed in real-time.
The initial contract I signed in 2001 was for three weeks. As originally conceived, it was to be a fairly simple programming job not unlike other sample libraries that had been available up to that time. He sent me the edited samples to examine. I was amazed at the sheer bulk of what had been recorded. I told him that we certainly could do a simple library as originally planned or . . . we could do something much more ambitious and complex, given the massive amount of material we had available. We discussed the possibilities and decided to go for the larger concept. The three week contract became 5 months of grueling work which resulted in a ground-breaking, high-end professional string library, beyond anything previously offered commercially. Gary sweetened the pot for the customer by creating a lavish package with extensive documentation (both historical and operational) and even personalized the binder with the specific customer's name embossed in gold! First class, like no library had ever been first class before. It was also rarefied in its pricing for the time and required considerable computer resources to take full advantage of the library. The library was enthusiastically received by professionals and this brought magazine articles and awards, including our first Electronic Musician Magazine Editor's Choice Award. They actually created a new Editor's Choice category for sample libraries because of the product. Healthy sales followed and the library became a staple of major Hollywood composers. I became Garritan's Director of Programming for future products, a position I held (along with simultaneous additional positions later) until I retired.
For our next trick we decided to move in a completely different direction: Instead of an all-out high-end library limited to professionals Gary wanted to aim at Everyman with a product that would require minimal computer resources; have a small total sample footprint on the hard drive; contain strings, woodwinds, brass, and percussion plus some extra “goodies;” and be priced at a level that any project studio could afford – a fraction of the price of the Garritan Orchestral Strings. We also moved it to a new platform: The Native Instruments' Kontakt Player format, to help make it simple to operate and inexpensive to use because the player came with the library, eliminating the need to buy separate sampling software. Native Instruments is a German company, which resulted in my work hours being extended into the wee hours, close to dawn, in case I had to call and talk with them at the beginning of their European work day. "Them" being excellent people like Martin Jann, by the way. Anyway, it fell to me to find ways of accomplishing this with as few sound quality compromises as possible given the difficult goals. I developed a system of MIDI controllers that used a variety of programming techniques to simulate behavior that normally required separate samples. In other words, I utilized MIDI controllers to dynamically manipulate various aspects of the sound (attack strength, decay time, dynamic brightness, vibrato, legato behavior, note variability, portamento, etc.) and accomplish effective results with minimal sample material. We also introduced a concept of "ensemble building" that gave great flexibility in creating groups of instruments of almost any combination and size. This library was also well received and by a much larger audience since almost anyone could afford it. This was the Garritan Personal Orchestra (GPO) and it became one of the most successful sample libraries in history, easily out-selling the Giovanni Gabrieli sackbut library of 1727. As you can imagine, total memory in the 1700s was limited to the capacity of a pen quill. But enough dry history.
The Kontakt Sampler
Native Instruments' powerful Kontakt platform acted as the format for several of our libraries. To the right is a view of the programming editor interface where I did the instrument design work during those years. The structure of the instrument is in modular form as can be seen on the right. The challenge is to analyze each of the constituent elements that gives an instrument its intrinsic identity - attack, sustain, decay, dynamic harmonic structure, note transition behavior, extraneous noise characteristics, etc. These elements are each treated separately and many of them given controllers that the user can manipulate as needed. Other important characteristics aid in the illusion by supplying such things as note-to-note intonational and timbral variability. When building ensembles with separate instruments steps needed to be taken to avoid phasing problems during unison passages. Tools were also suppled to place the instruments in a believable acoustic space. We took advantage of advances in sampler technology to add new features to the products through updates. Most updates were suppled free of charge to users.
A successful virtual instrument is a balancing act of a multitude of elements in the most effective proportions. Finally, the user of a virtual instrument must possess intimate knowledge of the performance capabilities and idiosyncrasies of a given instrument to use the programming tools to greatest advantage. Now, for an injection of stark reality: No virtual instrument totally captures all of the characteristics of an acoustic instrument because of the massive variabilities that exist from moment to moment during live performance. The best one can hope for (under ideal conditions) is something that strongly resembles a real acoustic instrument much of the time.
Garritan Personal Orchestra was delivered in a software player like the one on the right. Up to eight instruments could be loaded per player "instance." Certain keys on the keyboard outside the actual range of the instrument were appropriated to switch between different performance techniques. Those keys were displayed in yellow while the range of the instrument was displayed in blue. The violin section to the right could be switched between sustained long bows, short bows, alternating up and down bows, pizzicato, tremolo and trills. All except pizzicato were available in both standard and muted versions.
Advanced sampling was joined with modeling techniques to produce the solo Stradivarius Violin and solo Gofriller Cello. These were designed by Dr. Giorgio Tommasini and Stefano Lucato and were part of the Garritan line for several years. Giorgio and Stefano went on to join forces with Peter Siedlaczek to create their own company named Samplemodeling. The fruits of their work can be heard on the EWI page of this site. They have extended sample/modeling techniques to woodwinds and brass instruments with impressive results.
We followed GPO with the library that was more my baby than any other: The Garritan Jazz and Big Band library (JABB.) In this one I extended the idea of controllers to more areas of sound emulation since Jazz typically employs a larger palette of unusual instrumental performance techniques in standard practice - growls, doits, falloffs, "kiss" releases, etc. For one thing, I found ways of creating convincing and fully-controllable trumpet “shakes” without the need for actual recordings of shakes. The library contained a comprehensive collection of saxophones, lots of brass mute choices, extended instrument ranges, rhythm section instruments, completely programmable brush work for the drums, etc. More awards were garnered from many publications including another EM magazine Editor's Choice Award. This also forced a normally respectable person like EM editor Steve Oppenheimer to have his picture taken with me (see below.) Here it is years later and I suspect he's still embarrassed at the association.
Receiving the Electronic Musician Magazine Editor's Choice Award for JABB. Gary Garritan, editor Steve Oppenheimer, Tom - Los Angeles, 2007.
Additional libraries followed including the Garritan Concert And Marching Band but there were other events as well. We traveled, and traveled, and traveled some more. Every January we attended the NAMM (National Association of Music Merchants) show in Los Angeles where we would setup a couple of booths for product demonstrations while Gary would have meetings with distributors and others – he had an extraordinary talent for business deals! He was (and is) the most brilliant businessman I have ever encountered. He could come away from a NAMM show with 20 deals in his pocket - and, to my consternation, a truck load of new work for me to tackle. His business acumen is definitely a talent for which I have neither aptitude nor interest.
The NAMM (National Association of Music Merchants) show in Anaheim, CA. Starting at the upper left moving clockwise: 1.) Gary directing the setup of the main booth; 2.) Jeff Hurchalla setting up the secondary booth which was used primarily for demonstrations of the libraries. Jim Ortner took the primary responsibilities for doing personal demonstrations of products.; 3.) Part of the Garritan programming team: Composer, conductor, violist, bassist, ex-trumpeter, programmer extraordinaire Jeannot Welter and math/programming wizard Jeff Hurchalla.; 4.) Tom discussing convolution techniques with expert Ernest Cholakis; 5.) Garth Hjelte of Chicken Systems and Jim Ortner; 6.) Jeff Hurchalla preparing for the demonstration of the Steinway piano library.
Sometimes we'd attend summer NAMM as well. Above on the left is a rare shot of me about to give a video interview for a webcast at the Indianapolis Summer NAMM in 2005 with Mike Salamone, me, and Gary. On the right the three of us (inspired by the Four Freshmen) on display as the Three Stalemen.
One of the perks of doing all of this was the opportunity to meet and talk with many prominent people within the music industry, including many of my favorite musicians. They freely offered compliments, criticisms, and suggestions. It was generally enlightening and entertaining. It included dinners, meetings, concerts and occasionally included invitations to visit the personal musical lairs of historically significant musicians. Some, like the amazing electronic music pioneer Wendy Carlos ("Switched On Bach,") became very good friends and generously contributed valuable insights and suggestions for our products. A few years ago Wendy even created a page on her website concerning all things Garritan. You'll find a very kind mention of me near the bottom of this page.
Left photo: Gary, me, and Wendy Carlos at the AES (Audio Engineering Society) convention in New York City. Middle photo: Manning the Garritan booth with legendary jazz bassist Chuck Israels at the International Association of Jazz Educators in New York City - 2007. I also delivered a presentation on the use of virtual instruments in music composition at the IAJE that year. Right photo: Discussing music technology with jazz great Herbie Hancock at NAMM in L.A.
I occasionally gave audio-visual presentations on the design and application of virtual instruments in general and the JABB library in particular. The photo on the right was taken during a presentation before the Seattle Composers Alliance in 2006 but I also gave presentations at the IAJE in New York and participated in presentations for the Society of Composers and Lyricists in Los Angeles.
It was always nice visiting New York again after I had spent time there in the '70s studying trumpet. Another memorable sojourn was a week-long trip to George Lucas' Skywalker Ranch to record a Steinway piano for a proposed library. A great place to work! See slideshow feature below.
The Garritan team grew rapidly from just Gary, his wife Marianne, and me to a larger operation with people specializing in various aspects of library and sample player design plus various aspects of business. It was quite a brilliant assemblage of programming talent including Jeannot Welter, Jeff Hurchalla, David Viens, Markleford Friedman, and others. We eventually moved from Native Instruments' Kontakt platform to our own ARIA sample playback engine, developed for us by Plogue Art et Technologie, Inc. Having our own player platform had the advantage of placing all aspects of product creation and production under Garritan control. It also increased the speed of our responsiveness to problems.
The Garritan Corporation, through David Viens' Plogue Art et Technologie, Inc., developed the ARIA player based on the sfz spec. The interface to the right shows 16 slots at the top for loading instruments, assigning MIDI channels, polyphony and outputs; a mixer below with volume sliders, pan and auxiliary send knobs, plus mute and solo buttons; and a keyboard display at the bottom to show instrument range and activity. Page tabs and pedals are shown to the right and a mod wheel to the left. Each tab opens a page devoted to various control specifics, room simulations, EQ, and settings. If more than 16 instruments are required additional instances of the ARIA player can be opened allowing for 16 additional instrument assignments per instance.
Garritan produced the first authorized Steinway piano sample library, shown on the right. It had many advanced features designed primarily by Jeff Hurchalla. Steinway and Sons supplied superb, hand-picked instruments for recording the samples along with providing first-rate technicians to maintain the instruments during recording. Every note at many different dynamic levels were recorded from the perspective of a wide variety of microphone arrangements. Sustain and sostenuto pedals were also part of the emulation. The result was a gorgeous sounding piano library with the Steinway stamp of approval.
Panorama of Benaroya Hall in Seattle where we recorded brass instruments and hall impulses by chair position for a proposed advanced convolution engine. In addition to studio recording we also recorded at many different halls from Seattle to New York, in Canada, and even the Czech Republic.
Other chores that went with the job were related to the community of Garritan customers. The most visible part was the Garritan forum which tended to have considerable traffic. I was one of the moderators and for years I would answer many of the technical questions as they arose from the usually bright and pleasant customers. It was also part of the job description to deal with the occasional insulting or disruptive poster - ah, the Internet. If the unpleasantness couldn't be dealt with in a reasonable manner then the last resort would be to send me in to dispatch said poster with withering sarcasm - one of my major talents! And also why I'm so lovable. Occasionally, unfortunate recipients of my dry humor would call for my immediate dismissal (an understandable reaction not without some justification.) I pleaded with Gary to accept this strongly-offered termination advice but he just refused to let me go. The fact that he had compromising photographs of me when I was a young, aspiring swimsuit model (desperately in need of cash) gave him all the leverage he needed to force me to stay. Drat!
Over the years, my responsibilities were expanded from Director of Programming to include Vice President of the Corporation and eventually President (when Marianne needed to take a temporary leave of absence) but the job description really didn't change that much from my perspective. By the way, I should mention that Marianne proved to be the rock behind the Garritan Corporation. She handled many of the more difficult chores of the business and was endlessly supportive of Gary and the Corporation. I don't think any of this would have worked well without her.
The photo to the right shows pilot Jasper "Blades" Appoppin, Gary, and Marianne during their "copter" project where they attempted to record and sample as many different models of helicopters as possible. The project arose out of an odd canard about Bell Helicopters producing a musical pitch when struck by a clapper - a "copter clapper." Now, this may actually be partially true but the pitch tends to be buried by the noise from the engine and blades. Besides, I could have saved them a lot of time and effort if they had asked me about it. I would have told them that most commercial aircraft, rather than producing a pitch when struck, make a sound similar to "Boeing." But enough of this Top Secret (sometimes spelled "bogus") information and apologies to Gary for revealing it on the Interwebs.
Anyway, that, boys and girls, is how a three week contract turned into a ten year gig, which ended with my retirement in June of 2011. Many thanks to Gary and Marianne for an extended career path I could not possibly have foreseen in the year 2000. To be fair, you should be aware that I can't even foresee things two weeks in advance.
Garritan Sample Library Demos
Garritan Orchestral Strings (GOS)
The following "GOS Demo" was an elaborate demonstration I prepared for the 2003 NAMM show in Los Angeles. It gives a fairly comprehensive overview of the various features of the Garritan Orchestral strings library. I get the narration chore and wrote both the text and most of the music demonstration material with the exception of the opening music which I would credit if I could just remember who did it. Oh well, too many years and missing documentation. I'll add it if I ever find out:
I wrote the following brief piece to show the quiet, legato side of the string library. This was used as the "Theme" for the Garritan Website for several years and was heard at the end of the NAMM demo above:
This is an excerpt from the beginning of the "3rd Movement of Rimsky-Korsakov's Scheherazade" showing GOS with some standard symphonic repertoire:
Garritan Personal Orchestra (GPO)
The best demo of GPO is probably the video demo on the GPO page at Makemusic. This demo shows an actual performance of a winning entry in a GPO contest; you see the live performance but you hear all GPO instruments in a replacement mix - it is stunningly effective and was produced by Dan Kury. Here's a link: Makemusic and GPO demos. Look down the linked page until you get to the "Knights and Magic" contest winner video clip. There are also many other audio demos on this page as well.
Here's one smaller ensemble demo where I did the final programming and mixing using solo instruments from GPO. It's a Scott Joplin ragtime tune called "Easy Winners." This track features violin, flute, clarinet, two trombones, tuba, piano, bass drum, snare drum, and cymbals - all from the GPO library:
Garritan Jazz and Big Band (JABB)
"Big Band Demo" (5 saxes, 8 brass) was composed by Gary Lindsay for his book Jazz Arranging Techniques. I used the notation from the book as a reference in creating the MIDI sequenced demo using all JABB instruments:
"Margot's Mood" was composed and arranged by Chuck Israels. Again, I used music notation as my reference to create this MIDI sequenced mockup using JABB instruments:
The next is an illustration of the brush drum kit where the brush stirs are fully controllable using Aftertouch data. Because the stirs are controlled by data they are independent of tempo and will actually follow changes of tempo as in this demo:
I composed "Psychosis in Trumpetsville" to illustrate JABB's high note trumpet capabilities. Again, this arrangement uses only the instruments in the Jazz and Big Band sample library:
Trumpet High Notes - the file below serves as a peculiar addendum to the "Psychosis In Trumpetsville" demo and consists of a group of recordings primarily employed by me for comparison purposes during the programming phase of the trumpets for the Garritan Jazz & Big Band library. The first three are reference recordings for the unique “shakes” programming I developed for the trumpets and the fourth is the actual recording of the highest note used in the library – a triple high C. All notes are recorded as played by me; no audio editing trickery was used in this stark illustration of "Trumpet Player's Disease." This was my final recording session as a trumpet player and the photo to the right was taken about the time of those recordings. Without a doubt I went out on a high note:
Techniques for Demo Creation
When I created the demos above I used a number of procedures to achieve reasonably natural sounding acoustic emulations. I thought I'd briefly mention a few here (this could be a book!):
- All the demos were created within a sequencing environment (I mostly used Sonar on PC) so that I could easily view and manipulate the note and MIDI controller data to "sculpt" the parts one by one. Each track would contain default data for volume/timbre at the beginning. This would be refined in detail, note-by-note, later. I used the "piano roll" view for most things.
- I would begin by recording primary parts (1st violins, drums, lead trumpet, etc.) Inner parts would come later using the primary parts as a reference.
- I initially played each part separately from a keyboard for musical interpretation of timing and note values and to make certain that no two parts were exactly the same, just as no two parts are played exactly the same by musicians in a live performance. Nothing was ever quantized - an essential rule! Phrasing was a part of this step - remember to leave room for horn players to breathe!
- Once the initial recording of note data was complete I would then begin refining the timing and note value data and enter articulation information.
- Next, I would draw in the cc1 or cc11 data (choose one) for the shaping of volume/timbre note-by-note.
- I usually chose to draw in the vibrato data next (if applicable.)
- Finally, I would refine the details of attacks, legato smoothness, timbre and intonational variability, scoops, shakes, falloffs, etc., as needed, using the designed controllers for each. Of course, specifics depended upon the instrument type.
- Then I would proceed with recording the next part. And so on until all parts were recorded and preliminary editing complete.
- As soon as all parts were "roughed in" I would do a preliminary mix to place them in their seating positions and listen carefully to the interaction of the players and make adjustments to each part, as necessary.
- Once all the parts were edited to my satisfaction (translation: "I'm out of time, that will have to do") then it was time for the final mixing using a combination of panning and layered convolution reverb. "Layered" means using more than one impulse for each depth position on the stage. In the case of the big band demos, that meant three: one for the saxes in front, one for the trombones and rhythm section mid back, and one for the trumpets in the back. I tended to use between one and three for orchestral music.
- Finally, if I were doing these demos today I would make one change in my procedure for most of the parts: I would record the initial data with the EWI instead of the keyboard. I'm a much better EWI player than a keyboard player! It would also allow me to enter note data, volume/timbre data, vibrato data, and articulation data (to a point) in one pass. A time saver.
One of the least easily communicated parts of the process of entering controller data is how decisions are made for adding each type of data. Alas, the answer is: I use my musical intuition acquired over a lifetime of listening and performing. I know what I would do if I were playing a particular part and have a pretty good idea of what other professional musicians would do too. There is no easy shortcut around learning these things - if you have the professional playing experience, great: Use it to make decisions. If not, listen to lots of recordings of great professionals and when you are done . . . listen some more. Close, detailed, analytical listening. A good practice technique is to find a recording of a great performance and import the recording to an audio track in a sequencer. Then try to match every detail with your virtual instruments on parallel MIDI tracks. The more you try, the more you will learn.
When Sample Libraries Are More Than Just Sample Libraries
Sometimes technical projects intended for one purpose lend themselves to other unforeseen purposes. One example creating a great deal of personal satisfaction is the use of Garritan sample libraries for the Drake Music Project Scotland. Follow the link to Gary's explanation with photos and audio/video files.
The Garritan team spent a very enjoyable week in 2009 at George Lucas' Skywalker Ranch recording a second Steinway piano for a future library. We used the main scoring stage where the soundtracks to many movies have been recorded. The piano was setup mid soundstage and surrounded by a forest of microphones and diffuser panels. Here are some souvenir photos from that week: