Making orchestral music
...without an orchestra

The Phoenix Project
By Bjorn Lynne



The first thing I must do is to explain that this article is meant for the average music listener and people who are curious about music production and arrangement. If you are already an experienced composer or arranger of music, this will teach you nothing new. I am not a classically trained or educated orchestrator. If you are one yourself, you probably know how to do all this stuff much better than I do. I'm simply telling the story of how I carried out my first orchestral project. I'm not saying it's the only way. In fact, I'm not even saying it's the right way. I'm just saying it's the way I did it, and people around me were happy with the results. This was written simply as an amusing little story about how I put together the orchestral music without an orchestra. The whole thing is for entertainment - not for reference.

Tools at hand

I had the following equipment at my disposal for this project:

* Kurzweil K2500 synth/sampler with 64 mb ram.
* Soundblaster Live! soundcard
* Yamaha VL-70m synth with breath controller
* Novation Supernova synth
* Novation BassStation synth
* Peter Siedlaczek "Advanced Orchestra" sample-CDs (audio versions)
* Mixer and various outboard reverb units etc.

Sticking to the instruments

I knew from the start that I set out to write about 1 hour's worth of music. Some big and powerful - some calm, timid and some dreamy. Well, actually, let me correct that. I wanted to write enough music to make it a one-hour project in the end. I always throw away some bits and pieces before the final stage, so with that in mind, I figured I had to write about an hour and a half worth of music.

The first thing I learned was that it's important to set up a collection of instruments in the synths and samplers, and then sticking to it throughout the whole project. So the first thing I did was to spend about 2 weeks just working samples. I had bought the audio-versions of the sample-CDs with orchestral sounds. The same CDs were also available (a lot more expensively!) in "ready-to-use" CDROM format, ready to load into the sampler and just play. But the audio-versions of the same CDs were a lot cheaper, so I did the rather tiresome job of sampling each sound from the CD, and put it into my sampler at the right place. Each instrument was given a new sample for every 3 keys up the keyboard, i.e. C, D#, F#, and A for each octave on the keyboard. Over 5 octaves, that makes 20 samples for a single instrument. And each sample had to be properly looped, mapped, and so on. So it was quite a tiresome and longwinded job. But I think the fact that I did it myself made me a little bit more "aware" of each instrument and each sound, and I had a better relationship with them, rather than having just loaded them up in two seconds. It gave me a better feeling of what made each instrument "tick".

Back to what I was saying about creating a basic set of instrument and sticking to it throughout the project. It's quite obvious. The orchestra doesn't change during recording of a whole 1-hour piece. The performers don't move around, or replace instruments. So I created a set of about 30 different instrument, saved all the files, and decided to use them for the whole project.

I didn't want a "pure" orchestral sound. I wanted a "basically orchestral sound, but with other sounds also mixed in from time to time". These "various other" sounds were going to be mostly analogue synths playing "other-world'ly" odd sounds that are not supposed to sound like any instruments, but are just "sounds". So I made up a basic "skeleton" of orchestral sounds, and decided to use that as the back bone of the compositions, and then throw other "weird" sounds into it whenever I felt like it. If the "weird" sounds changed from piece to piece, but the orchestral backbone stayed the same, I figured that would give a nice balance of continuity versus variation. So the most important thing was to first get a good set of backbone orchestral sounds to build on.

Staying in control

All the brass samples were stored in the Kurzweil K2500. The strings were stored in the PC's memory, and played out through the Soundblaster Live's digital output. Brass and strings. The two most important and dominant families of instruments in an orchestra - at least in sci-fi and space-theme music. Brass consisting of separate tuba, trombone, french horn and trumpet samples. Strings consisting of separate bass, celli, viola and violins. All the brass- and string sounds were taken from the Peter Siedlaczek Advanced Orchestra sample CD series. But I "built" the instruments myself, to my taste. I wanted to get some real-time control of the sounds, so that I could put some colour to them as I played them on the keyboard. So I applied a low-pass filter to the brass sounds and set up the K2500 so that I controlled that low pass filter with a slider on my keyboard. This allowed me to "open" and "close" the brass sound as I went along, to create "swells" and "stabs" while playing the parts. Rather than just play a flat brass sample, I could then add some more emotion to it, giving the impression of some notes being played softly, others hard.

MP3 sample: brass sample simply being played "straight"
MP3 sample: brass sample played with real time control of the low-pass filter

Likewise with the strings, I wanted some real time control - something that I could use to "colour" the sounds as I played them. It was important to me to be able to change the sounds during recording, because the notes and the way they are performed interact so much with each other. It just wouldn't have been the same if I had played all the notes straight out first, and then applied the variations in filter, attack-time and so on. I decided that it had to be done together. So with the strings, I applied a slider on the keyboard that let me control the attack- and release-times, i.e. how fast the note rises to full volume, and how fast it drops back down to silence.

MP3 sample: violins played with fast attack/release times
MP3 sample: violins played with longer attack/release times
MP3 sample: violins played with very long attack/release times

Placing the "virtual musicians"

In an orchestra, each performer has his place. I know that some people, when they want to "emulate" an orchestra in the way that I did, go to almost extreme lengths of getting everything correct right down to the last millimeter. There are huge books available on how to orchestrate music. A certain type of 18th century Spanish violin, for example, could not play the same high A as a 20th century one. And so on.

However, I decided to go a bit easy on all this, stick to the basics, and use a bit of common sense. The two arguments I used were (1) no music will ever be exact science and (2) I just haven't got time to read 2000 pages about the way different violins work. I did read a quick guide to the tonal range of each instrument, and a rough guide to the placement of each instrument in a concert setting. I used this to draw the following sketch:

(This is my actual sketch as used during setup. I'm sorry that it's a bit messy - I didn't think that anybody besides myself would ever see it!) :-)

Using this sketch, I set up the volumes and left-right position of each instrument on my synths and inside the music software (Cakewalk Pro Audio). I also set up the amounts of reverb added to each instruments so that the instruments furthest away from the listener had more reverb, and the closer instruments had less reverb and more of a direct sound.

Playing the parts

Again, I decided right from the start that I wanted to play all the parts myself on the keyboard. It is possible to "tap in" and "type in" the music into the music-software, but I decided to stick with playing every little bit to try and keep it "human".

For someone (like me) who has been playing pop, rock and other types of music for many years, it just seems all too tempting to follow the same basic ways of playing. If you want a Cminor7, you slap 4 fingers on the keyboard - C, Eb, G and B - all done, no problemo. You can't do this in orchestral music. That's not how a chord is played by an orchestra. Instead, the celli would play the C, the viola would play the Eb and the G, and the violins would play the G. For example. And it would all come out in harmony (hopefully). Furthermore, the instrument would rarely just play that note and stop. It would be part of a sequence. The chords could be inverted for the next passage. This turned out to be harder than I expected because I was so used to the habits of pop/rock music that I hardly managed to shake it off and think in a different way.

Again, I decided not to go completely nuts about realism. A true fanatic would surely find out about which direction the second violinist would draw his bow over the strings first, and how much vibrato he would put on each note. But again I decided to go with a bit of common sense, and try to use my ears. I tried to play phrases on each instrument that would be possible. The trombone does not play for 45 seconds without drawing breath - and so on. A kind of middle ground between a little bit of realism and a little bit of just trying to make it sound good. That was the recipe I had used throughout the whole project, and it seemed to serve me well again.

Catching some breath

What I enjoyed most of all was to play the woodwind parts. I almost wish there had been more of them. But this was a sci-fi / space theme and because of that, brass and strings were always going to be the most dominant instruments. For the woodwinds, I used the breath-controller and actually played the notes in the part on the keyboard, while at the same time "blowing" the parts into the breath controller. This gave me a more realistic sound and when put into context with the rest of the orchestra, I felt that the woodwinds sounded really good. Maybe it was just because I knew that I had literally blown the notes, that I had a special feeling about those few notes of woodwind. :-)

Recording and Mixing

For the recording and post production, again, continuity was important. I set up my recording gear (a DAT recorder and some outboard effect units) and saved all the setups so that I could record all the pieces with the same type of sound. I'm afraid I didn't succeed 100% with this, because of some occasional slackness on my part, some parts were recorded with slightly different settings. But again back to my point about not being too fanatical or hysterically correct about any of this. So all the music was recorded with roughly and basically the same settings. I then plugged the digital output of the DAT recorder back into the PC equipped with a Frontier Design WaveCenter digital in/out card, and put all the music digitally onto the PCs harddrive, from where I balanced the volumes and applied the final compression and EQ.

With pop and rock music it's a common thing to add a lot of "compression" to the audio level (that's volume-level compression, not file-size compression) in order to make the final mix sound "louder" and more "radio friendly". But for this music I wanted to try and keep a lot of the original dynamics - and keep the quiet parts and the loud parts well separated dynamically. So I only applied a little light compression, and practically no EQ.

And that was it! My first orchestral project - without an orchestra. The whole thing took me about 7-8 months, although I was also doing some other things in between during that period. I had a great time doing it, and I hope to be able to do it again.

- Bjorn Lynne
October 1999


All text, music, photos copyright Bjørn Lynne.