Sooner or later, we're going to need to some effects in our library. Tonight I'm looking at (listening to) the DX7 REFS WHISL.
First, let's get one thing straight: Music/sound is about LISTENING. That's why I always hesitate to say that I'm "looking" at a patch. I think we have to be careful that we listen because of all the parameters, and I mean hard data, that's right there in front of us on the UI. I'd love to live in a world where there are no NUMBERS on our synths. I also dig rotary controls--no upper/lower limit, we just keep on rollin'. ANYTHING, any opportunity to just listen is the heart and soul of music.
I teach piano lessons privately. Young and old, I tell them the same thing: The stuff you see in front of you is nothing but ink and paper. Music is what you hear. It's communicating in some kind of meaningful way, even if the meaning can't quite be put into words. I think of music more often as that which can't be said. It's a lot of things to a lot of people. I just tend to dwell on the more abstract.
I just hate looking at sound and music as just a bunch of numbers. Unfortunately, the GUIs often communicate with us in numerical form. It doesn't help that the first synths seem to have been developed by mathematicians and scientists. There are some advantages, though, because it allows us to converse on the topic in concrete terms.
Back to the patch:
I'm looking at (yes, looking--sorry) REFS WHISL in FM8. What stands out about this patch? If I listen to one note--well, it sounds like what you'd expect. Play another note, and--whoa... Exactly the same. OK, now actually looking at it, I can see that it's as I'd guess with no tuning ratios. Everything is in absolute, concrete numbers, which means no absolute, concrete harmonic relationships.
Let's look at Ops A, B, C, and F since this appears to be the most sonically complex part of the patch. Op A is 10Hz, Op B is 1Hz. This looks more like an LFO than anything useful by itself. Now we have to think about what LFOs are used for.
Most often LFOs are good for pitch variation (vibrato, trills, usually controlled with a mod wheel to change depth or width of the variation). LFOs are used for other things, of course, but in beginning sound design, pitch is a good target. So it appears that we're using Ops A and B as LFOs of each other. It's FREQUENCY modulation, right?
Well, there's a slight problem with this line of thinking. While frequency variation, whether with LFO or FM modulators, makes for some interesting timbre variation, FM on the DX7 isn't frequency modulation in the truest sense of the term. There is some variation in pitch, but not really that much. This becomes very evident when using Thor to copy DX7 patches because using one oscillator to mod the pitch of another causes extreme variance in pitch/timbre. There's so much variance that the resultant pitch and sidebands are horribly unstable, which explains why feedback loops in Thor don't really work very well. Why is it FM works so well on the DX7 when it sucks on the more advanced Thor?
The answer is that the DX7 doesn't really change frequency, at least not very much. What happens on the DX7 is that, while there is some apparently detectable change in pitch, it's the resultant change in PHASE that is the true power of DX7 FM. Putting all that energy into altering the phases of sine waves allows for controllable sidebands that add a distinctive coloration to the DX7 timbre. I really admire the DX7 for this because adding just a touch of feedback to an operator or chain of operators gives the DX an almost analog quality. This really comes through on brass patches!
So what does that mean if Thor's brand of FM is pitch unstable when compared to the DX7? Quite simply, if you increase the amount of modulation one oscillator has on another, the resultant pitch is higher than you expect. In a future post, I'll explain how tuning--careful tuning--can compensate for this unfortunate side-effect and get more authentic FM sounds.
Meanwhile, we're going to pretend we don't know any of this and that Thor is exactly the same as the DX7 (you can stop laughing now).
Our first task is to get an FM pair in Thor to behave like not one but--wait for it--two LFOs, one modulated by the other. Can it even be done? "I'm sorry, Dave, I can't do that."
"Open the pod bay door, HAL." Excuse me while I give this thing the finger.
OK, fine. Whatever. But let's go back to FM8 for a sec. Everything else falls within the audible range. All we need is two LFOs, and in doing so we can get very precise results. In Thor, set LFO2 about 10Hz and LFO1 1Hz. In the matrix router, set LFO2 as source and LFO1 Rate as destination. I'm just guessing here, but try 25 for an amount.
Now here's the part where you have to use your brain. Remember, Thor only understands harmonic relationships. Totally guessing again, I'm trying mod 11 and carrier 8 with LFO1 modulating Osc1FM frequency, amt 65 in the router. Pretty wild, right? I'm checking this against FM8 and it's way off. I'm changing the routing to Osc1 pitch and backing the amount to 10. It still doesn't sound like ABC in FM8. What we need now is some envelope modulation. Set the mod EG decay time to 45.2 (keep it short). Remember how we used LFO2 to mod LFO1 rate? Now scale this with the mod EG. Crank it up to 100. Doesn't really do much does it? Set Filter EG decay time to 316 ms and kill the sustain. Use the Filter EG to scale LFO1 modulation of OSC1 Pitch. Set the amount to 100. If you play a lot of rapid notes, it almost sounds like a triangle roll.
Of course, the differences are so striking between Thor and the DX7 that it's impossible to pull this off without stopping and experimenting along the way. I changed the Osc1 carrier to 7 and FM amt. to 22. Getting better.
So far we've only used Osc1. Sounds like a great high-pitched metallic bar kind of sound. A ref's whistle is high pitched, right? Now for the rapid pitch and volume variation associate with a whistle sound.
To my ears, the most significant thing about it is the indeterminant aspect of this sound. Turn off the keyboard scaling for Osc2 and route Osc2 to Osc1 pitch. Make sure FM amount is all the way down, also. Turn the octave all the way down. Here's where the ear kicks in. Important point to remember here is that the DX7 FM effect is very subtle--change phase, not pitch. So for the amount of modulation for Osc2 to Osc1 pitch, set it to 15. The modulation is a bit slow for my taste. Speed it up just slightly by setting Osc2 semi to 6.
Almost there. Repeat the above procedure for Osc3. This time set semi to 2 and amount (in the router) to 39. So close!
The problem we have now is that the bandwidth is fairly broad within our hearing range. This means higher pitches are pushed out of hearing range and the lower end tends to dominate. Now we have to compensate by altering the fundamental pitch of Osc1. Set octave to 4, semi to 9 and tune to -12. Getting closer. Now the trouble is there's too much noise. Here's where filters come in. Set Filter 1 to State Variable, use the band pass filter, freq to 3.28 kHz, res at 123. Not perfect, but it's there.
I also did a little more tweaking in the router. Osc2 amount is 12 and Osc3 amount is 26. Done!
This is a great effect for a lot of different reasons. Obviously, if you NEED a whistle sound, there ya go. But you can get different effect by playing around with the FM amount and mod frequencies in Osc 2 and 3. You can make some great cricket and cicada sounds as well, or play around with the amp envelope to get different kinds of shakers or clackers.
Also keep in mind that part of what we're doing here is a study in how FM works, especially in Thor, and I'm a long way from being an expert. It's a great accomplishment, though. You really can't ignore the possibilities in the software for programming effects. I think it's good practice to work on these, especially given the difficulty of such a task.
No comments:
Post a Comment