|
TECHNIQUES FOR AIDING K2000 CHANNEL STEALING The K2000, although still a glorious synth, has only 24 voices of
polyphony unlike its newer brother the K2500 with 48 voices of
polyphony. What this means is that on a K2000 you can only have 24
notes sounding at the same time. After that, notes have to be
"stolen" to allow for new notes to sound. This is called "channel
stealing". The K2000 and K2500 have advanced channel stealing
algorithms to make this as un-apparent as possible. Nevertheless,
you'll occasionally run into situations where channel stealing can
still be heard. MIDI sequences with lots of notes, programs with
long, held-out samples such as drums, and stereo samples will all
push the limits of the K2000 voice structure pretty hard. There
are a few techniques, however, that may help! If you're a K2000
owner and are experiencing obvious channel stealing, try these
techniques.
1. Reduce
the number of tracks playing back on your K2000 at one time. If
you're working with a recorder, try recording some parts to tape or
hard disk so they don't have to all be playing back at one time on
your K2000. "Submixing" like this can go a long way toward reducing
the load on your K2000. 2. Stereo samples vs. mono samples. Stereo samples take up twice as many voices. So if you're working with all stereo samples, your K2000 now only has 12 notes of polyphony! Not good. If there are certain samples that you can get away with being mono, do it. Kick drums, snares and basses for example can often be mono and in a mix they'll still sound great. To make a stereo layer mono simply do the following:
3. The channel stealer works by
assigning 'priorities' to certain notes. If a note has a high
priority, it won't be stolen as easily. If a note has a low
priority, it will likely be one of the first notes to be stolen.
There is no parameter that allows you to assign priority to a note
or a MIDI channel unfortunately. But knowing how the stealer
algorithm assigns priorities can help you make programming choices
to influence the priorities of certain layers.
The K2000 channel stealing algorithm
only looks at the (Course) volume in the F4 Amp page and the AmpEnv
level to determine priority for stealing. If the note is louder
initially (determined at the time the note starts by the F4Amp
Course value), it has a higher priority and therefore won't be
stolen as easily and visa versa. So, for example, if the F4 Amp
Course volume is set to 12db on one layer and 6db on another layer,
the 12db layer will have a higher priority and not get stolen as
easily. Likewise, the 6db layer will be considered to be softer and
therefore get stolen sooner.
If you want to increase the
priority of a certain layer so that it doesn't get stolen as easily,
you should first try to increase the Course volume of the F4 Amp
page. Since you're not going to want that layer to actually sound
louder in the mix, you'll need to compensate elsewhere for that
volume increase. The best way to do this is to Pad the volume
before F4 Amp. To do this go into a previous block, such as F1,
and lower the Pad parameter level the same amount you are increasing
the volume in the F4 page. So for example, if you increase the F4
Amp Course level 6dB, you should set the F1 Pad parameter to -6dB.
6-18db should be enough to influence the stealer.
If you want to decrease the
priority of a certain layer so that it gets stolen easier (and
consequently other layers won't get stolen as easily) you should
first try to lower the Course volume on the F4 Amp page. Since
you're not going to want that layer to actually sound softer in the
mix, you'll need to compensate elsewhere. The best way to do this is
to increase the volume in the Output page. To do this go to the
Output page and add the same amount you are subtracting in the F4
page. So for example, if you decrease the F4 Amp Course level -6dB,
you should set the F1 Pad parameter to +6dB. 6-18db should be enough
to influence the stealer.
If you're working with programs that use
the non-linear algorithms such as DIST, SHAPER or GAIN, you'll want
to make sure you use those algorithms in a block prior to padding if
you're trying to increase priority.
4. Similarly,
the engine looks at the settings of the AmpEnv to determine the
relative level of how loud or soft a note is. If the AmpEnv decays
very quickly to zero, that note is determined to be softer sooner
and therefore open to stealing sooner. If, on the other hand, the
AmpEnv sustains at full volume over a long length of time (like a
gated drum envelope), that note is determined to be louder and
therefore not a high priority for stealing. In the case of User
(RAM) drums, this unfortunately is exactly what you DON'T want the
stealer to determine. Why not? Read on.
User drums, as appose to ROM drums which
are compressed at the factory, have a typical decaying envelope that
is natural to the drum sample. You hit a cymbal and it starts out
loud and then decays over some number of seconds. When you load this
sample into the K2000 you want it to play all the way through the
end of the sample just like the cymbal does. However, the default
Amplitude Envelope will only play all the way through the end of the
sample when you hit the key and hold it down, not when you hit the
key and release it. To accomplish this, you have to program a User
AmpEnv. In order for the sound, in our example the cymbal, to not
decay prematurely you need to program the User AmpEnv as a "gated
envelope", that is one that goes to 100% level in 0 seconds and
stays at a 100% level until the end of the sample. You would do this
for both the decay (when you're holding the key down) and the
release segments (when you hit and release the key.) Here's where
the channel stealing problem comes in. The stealer algorithm looks
at that AmpEnv and sees that it's full volume for some number of
seconds, probably 5-10 seconds in the case of a cymbal, and
therefore it assigns it a high priority for not being stolen. In
reality, however, this cymbal is audibly decaying and would
be a perfect candidate for stealing unobtrusively. We WANT that
cymbal to get stolen at a point when it's audibly lower in volume.
If it gets stolen when it's softer, you won't notice it and it opens
up a note channel for another audibly louder sound. So, what this
means is that we have to program the AmpEnv to decay so the stealer
knows to go ahead and steal this note when it audibly gets softer.
But in order to keep the full sound of the cymbal ringing out we
need to make up for that decaying envelope elsewhere in the
programming. Here's how:
Caution -Do this at your own
risk! You will be editing with some very loud volumes so turn your
speakers WAY down until you know what you're doing. Seriously! A. Select a layer that has a long decaying envelope such as a cymbal or tom layer. B. Go into the Program editor.
C.
Go to the AMPENV page and program a
decaying envelope over say 5 seconds using these settings:
(afterwards you'll want to adjust the time to fit the length of your
sample)
Att1: 0s and 100%, Att2: 0s and 0%,
Att3: 0s and 0%
D.
Go to the ENV2 page and program a volume
ramp envelope using these settings: (afterwards you'll want to
adjust the time to fit the length of your sample)
Att1: 5.00s and 100%, Att2:
0s and 0%, Att3: 0s and 0%,
E.
Go to the F4Amp page and set Source1 to
ENV2, Depth at +96db.
F. Slowly
turn up your K2000 while playing to make sure it worked. You should
not hear anything jumping out at you or blowing your ears apart. At
this time you can adjust the times of your envelopes to match the
length of your sample or until it sounds smooth. In you want a more
compressed sounding tom, here's the place to shorten the envelope
decay to do that. G. If you bypass the F4Amp setting (by setting the Depth to 0 temporarily) you can hear what the channel stealer is going to hear which should be a fairly fast decaying amplitude envelope. My suggestion is do to this trick to ALL layers that have a long decaying envelope. In the case of drums you'd want to program this for ALL your cymbals layers and ALL your toms layers at least. Good luck! |