View Full Version : Composing strings with expression controller...

Scott Cairns
04-27-2003, 09:23 PM
Hi everyone, hopefully this wont sound like too silly a question, but what is the best way to record expression events for a string section? I.E. track by track or the whole section in one go?

It occurred to me that when a conductor is controlling the rise and fall in volume the whole string section responds together as one voice.

When I try to do a string line with expression one by one; 1st V, 2nd V, Viola, Cello, Bass, I just dont get that cohesive rise and fall, the kind of unified swell that comes from a string section playing tight.

What does everyone else do?

Thanks, Scott.

04-28-2003, 10:38 AM
Interesting thought, I am definitely going to give this a try.

In Logic (which I use) it should be very simple to copy the mod wheel events (or expression events) from one part to another.

For sections that really should have the same dynamics envelopes, you\'d only have to record the mod wheel behavior once.

Curious if someone on the forum has applied this approach.

04-28-2003, 01:25 PM
This is a very good question.

The thing is that the rise and fall in volume in sections is rarely one \"envelope\" thing. It may sound like it but each person in the section respond differently, some people can produce more volume than others etc.

So I would think that the best way is to have multiple envelopes, however, do not have them very different. They must be (very?) similar rather than identical.

04-28-2003, 01:54 PM
Maybe play the first (say Vlns 1) as good as possible and then, as a shortcut, draw similar expression curves for the other groups. Just thinking aloud images/icons/smile.gif

I am not sure yet if simple copying of the curve would be that bad, because all groups (e.g. GOS Leg EXP patches) are often quite distinctly played, recorded and programmed. I don\'t think that re-using the same curve data would be too obvious.

Again, nice stuff to experiment with.

04-28-2003, 04:34 PM
For what it\'s worth, I seem to achieve the best results by performing each string section (line) separately, resulting in a unique MIDI expression \'curve\' for each string section. I try to get each section to sound natural in and of itself. If you perform the parts using a breath controller, as I do, this happens naturally anyway.

Having achieved a satisfactory MIDI performance, I then print each section to audio, one per track in my DAW. After balancing the string sections volume wise, I run the entire \'string orchestra\' through its own dedicated bus, so that I can add further dynamics to the strings alone (volume automation in the DAW), and also to balance the strings against the rest of the orchestra. Scott, this single volume fader for all of the strings take care of your concern about everyone responding together (in case the individual parts didn\'t make it happen on their own).

If the objective is maximum realism, and the strings are prominent in the mix, I think separate expression curves for each string instrument section will be necessary. Two obvious reasons come to mind. First, no two instruments respond in exactly the same way. And secondly, the difference in performance adds a variability from instrument to instrument (section to section) that may add a hint of realism to the overall mix. At least that\'s how I see it.


Scott Cairns
04-28-2003, 08:14 PM
Thanks for your replies everyone. As I have a Roland controller it doesnt come with a mod wheel. I have been using a data entry slider for expression control and it doesnt seem to be too accurate. This is why I was getting so much variation in each string part when I performed the expression control one at a time.

I actually have this (\"http://www.evolution.co.uk/moreinfo/hardware/uc33.htm\") controller on order so hopefully my CC\'s will be more accurate when I get it. I can certainly see the logic of performing the expression part for each section. As long as they are all close they would likely sound more \"human\".

Regards, Scott.