Sunday, April 12, 2015

Getting Real Pt 9. - Forgotten Magic

In the last part of the series (part 8), I talked about compact composition, untangling notes, and understanding the semantics of the composition during reorchestration. It might be helpful to understand a bit more of what I mean by "semantics". I think the easiest way to do so is to understand the different ways that composers of olde tried to create space on the sound-stage.

It's hard to imagine, but in the time before CPUs were fast enough to automatically calculate complex algorithms that give space to a sound, you had to figure out how to do this yourself. By "space" I mean the effect that your notes live in an environment larger than the room you are listening to them in. The typical examples are echos and reverb, though other techniques exist. It turns out in old-fashioned sample based trackers, echos are one of the easiest techniques to replicate.

Consider this three note sequence:

Track 01
00 C-5 01 ..
01 ... .. ..
02 D-5 01 ..
03 ... .. ..
04 E-5 01 ..
05 ... .. ..

By itself, this sequence plays three eighth-notes, and in general, they'll sound rather flat. It sounds like they're playing in a closed room (and they probably are).

Note consider the following six note sequence:

Track 01  Track 02
00 C-5 01 ..  ... .. ..
01 ... .. ..  ... .. ..
02 D-5 01 ..  ... .. ..
03 ... .. ..  ... .. ..
04 E-5 01 ..  ... .. ..
05 ... .. ..  ... .. ..
06 ... .. ..  C-5 01 30
07 ... .. ..  ... .. ..
08 ... .. ..  D-5 01 30
09 ... .. ..  ... .. ..
10 ... .. ..  E-5 01 30
11 ... .. ..  ... .. ..
12 ... .. ..  ... .. ..

It's the same three notes, but now they're repeated again, but at a lower volume. This is a simple echo, and it does wonders to give space to music. Instead of feeling like it's being played in a closed-off room, suddenly it's being played in a cathedral, or in front of a massive mountain range. It creates space, space that you as the listener are temporarily transported to while listening to the song.

There's variations on this technique: playing the echo sooner or later can change the echo from a 60's playing-in-a-tin-bucket reverb to a wide-open mountain range. Adding more echos (at subsequently diminished volumes) can magnify the effect.

This really works best when you can dedicate lots of channels to the effect. Two, three or four delays give even better space (and better mimic expensive multi-tap delay units), but consume tons of channels. This technique is especially effective for huge leads and other synths. Purple Motion makes extensive use of this technique throughout his composition -- for example, the opening patterns 3 through 12 use 3 channels for the main background synth and its echos (he also compactly inserts drums, other synths and orchestra hits where he can).

An older more compact style was to simply interleave the delay into the same channel. You trade-off note completion for space and use less channels, this works especially well for very short notes. I don't remember Purple Motion using this in his section. But Skaven does in the "flying through the city" sequence (see Skaven's music, sequence 15-21, track 08).

Track 08
00 A-3 0D ..
01 ... .. 30
02 C-4 0D ..
03 A-3 0D 30
04 D#4 0D ..
05 C-4 0D 30
06 A-3 0D ..
07 D#4 0D 30
08 C-4 0D ..
09 A-3 0D 30
10 D#4 0D ..
11 C-4 0D 30
12 A-3 0D ..
13 D#4 0D 30

and so on

All of the notes with a volume of 30 (the 4th column) are echos. I believe that the first volume of 30, set on no note on position 01, is not meant to cut the note started on position 00, but simply the result of laziness, Skaven simply went down to every other tick and set the volume to 30 and it was "good enough" (TM). Regardless of the reason, this technique is also very common in tracked music, especially older tracked music where channels were at a premium. However, it's not just a matter of squeezing space-giving effects into limited channel-space, this technique also has a particular sound -- the echos also cut off the main notes, giving a kind of queer compressed sound to the technique.

You might be thinking back to Part 8, where I was also trying to untangle the notes and the result was effectively this second echo technique. In actuality, I tried to maintain a separation between these two techniques where I found them when possible. Often I ended up removing either kind of echo and using DSP effects to reintroduce them with fewer bytes -- remember, each note-track-volume-length combination produces new data that raises the entropy of the track.

In either case, I had to figure out which technique the composer was using, decide if I wanted to replicate it, remove the echo if needed (to save space) and reintroduce it if desired via DSP effect for the final version. I had to figure out if the composer was describing a large space that his music was supposed to exist in or a small space, and make sure I tried to reproduce that as faithfully as possible.

No comments:

Post a Comment