Monday, July 9, 2007

The Farmdale Symphony

For the past two weeks, I have had the pleasure of working with brilliant Hollywood music editor and musician Steve McCroskey on a project I have wanted to see completed for a decade.

What do you do if you are an orchestral composer--and you have no orchestra? This can be a really difficult proposition. It's the proposition that my wife, composer Carol Worthey has wrestled with since I've known her. The answer is (of course): get an orchestra. Not a problem if you have Bill Gate's budget, simply hire a symphony. Most orchestral composers are usually associated with a university or music school or other academic connections where they have at least a testing ground for new works. Getting works performed can become an occupation in itself.

Years ago, as a computer expert, I knew there should be a way to duplicate a real orchestra utilizing computing power--and there was. The early attempts utilized synthesizers with limited numbers of channels and tape looping to re-create an orchestra (rather badly I might add). Later, in order to improve the quality of the speaking instrumental voices, "sampling" was invented. In this mode an actual instrument, say a violin, was recorded or "sampled" with an expert player playing single notes throughout the range of the that instrument. Each of these notes were then turned into individual pitched segments that could be "played" by the computer, thus permitting the computer to sound like a violin. Sampling now improved the way it sounded, but not by too much. There are nuances in each instrument, effects and performance conventions learned by human musicians but not particularly available to a computer musician. For example, the violin has pizzicato (plucking the strings with the fingertips), vibrato (a wavering of the tone produced by shaking the hand while holding a string), tremolando (a vibrating tone produced using the bow). If you want the sound of a real performer, then each of these effects have to be sampled as well for every note (pitch) in the instrument's range. This takes a lot of bandwidth.

Several other factors need consideration. Typically there are 25-30 different instruments that comprise an orchestra. Also, there are groupings within an instrument, e.g. Cello I, Cello II, Cello II. This means more channels and a need for more technical capability.

Technically, the ability to create an orchestra on a computer has paralleled the increase in computing speed, available disk space, available memory and available digital to analog (i.e. computer to audio) hardware interfaces. Early on, this not only was unsatisfactory sounding but very expensive (hundreds of thousands of dollars). Fortunately at the same time the capability has increased, the cost has reduced.

There was a time, not long ago, when a number of low-budget Hollywood films used entirely computer generated film scores to save money. They saved money, but generated such awful public response and musician response that this soon stopped. Clearly the technology was not good enough to hack it. Today, not only has the technology improved, but live musicians are routinely mixed in with sampled tracks not only saving production bucks but producing a pretty good overall product.

Carol has a new symphonic work headed to China in 2008 recently completed. We really, really wanted to preview it--saving the world premiere for 2008. Carol, I might add, composes and hears her works in her head then notates the music directly to computer. She has done this since 1984 being a very early adopter of computer technology. Thanks to the help and support of concert pianist, Mary Au we connected to Steve McCroskey and the Farmdale Symphony.

Who are the Farmdale Symphony?

The more appropriate question is: "What is the Farmdale Symphony?" It is a computer piloted by a master editor running at the edge of affordable capability. Conducted by Steve, we transformed Carol's full orchestral score to an audio performance using sampled sounds. It sounds pretty darned good too! Not the LA Philharmonic as yet, but an acceptable performance not easily discernible as a computer performance.

I've wanted this capability for decades and at last it is real, do-able and within reach. I'm really pleased.

No comments: