Kurzweil:Dead Letters 1995b
From Sonikmatter
1995 Part 2
Drum Programming
Date: Fri, 8 Dec 1995
From: Sabirsh Al
Subject: programming: drum patches and hi/lo filters
Darcy Veer wrote:
For example, I like the "Real Drums" patch, but want my hihat and kick pretty dry and my snare real wet. How do I assign the different individual percusion instruments to different outputs? I hope this isn't too elementary.
Edit the drum program (press edit button). Then press the key that triggers the drum sound you want, while holding the enter button. This will take you to the relevent layer for that sound (assuming operating system ver. >= 2.x, I think). Then go to the output page and change the outputs from A(mix) to B(dry). The drum sounds within that layer should now come out the mix outputs without any effects - they will be "dry". AL
James wrote: I think I may have overlooked something, but how would I go about "stacking" a low pass _and_ a high pass filter onto the same one-layer program? Jim replied: Use a bandpass. A lo and hi in series effectively ARE a bandpass filter.
My .02$: A bandpass will work of course, but then you only get control over the central frequency and the width of the filter - or the distance between the peaks as with double peaked bandpass curves. If you want independent control of the width of the band (control over the low pass and high pass frequencies) then try using one if the alogorythms that contains both high and low pass filters (I can't say off hand what number that would be). Then you can control more precisely the cutoff frequencies of the two filters with real time controllers. AL
DX SAMPLING
Date: Wed, 17 May 1995
From: nursingh@CLYDE.ENS.INSA-RENNES.FR
Subject: DX sounds sampling
Mike,
>loop points
Ear Run. But the OS can help determining possible loop points if u press the - and + keys simultaneously.
> Do I have to use some sort of algorithm to beef up the sound?
The K2k is reported not to be transparent when it comes to samples.That is, when you sample into the K2k, the sample loses some low end and high end. However, this can be remedied using suitable DSP functions in appropriate algs. Assuming you've read Darwin's post about multi-sampling ... Once you have your multi-sample keymap, or multi-layer program or whatever, for tweaking you can try:
1. ALG 1 with the HIFREQ STIMULATOR. Now, much timbral information is contained in the attack of such samples. Correct dosing of the FREQ and DRV and AMP can give stunning results, especially in guitar and e-piano samples. This makes up for high-end loss. You can choose to route velocity to DRV or fREQ, following what you want to achieve.
2. ALG 1 with the STEEP BASS function. Pretty powerful filter. You can boost your low end overboard with this DSP function. You can try LOWPASS filters in other ALG's too. However, LOWPASS will remove top-end harmonics above the CUTOFF FREQ.
3. Layering a SINE with your sample. For this, you may have to look for a. A split-path alg. which enables a SINE to be played with your sample. b. An alg with SINE+.
Experiment with layering a very high-pitched SINE or a very Low-pitched one, with various differences in the realtive amplitudes of the SINE and sample (possible only if u use a split path or several layers)
4. Using algs with the +AMP and !AMP. These can do wonders with FM samples. Ear run, noramlly +AMP works best if u use a SINE (like in 3) with the sample.
5. Give LPCLIP a try. It's found in ALG >10, somewhere. It's particular but v. efficient for low-end.May induce distortion though.So use Pad to soften along the line.
6. On DX's, velocity seems to actually add harmonics to the tone instead of just opening a filter. Additive synthesis can be tried too with the WRAP function but dosed v.low and/or the SHAPER function( set to .250 or an integral multiple of it if following WARP , else set to suit your ear ).
7. Later on, you'll probably want to have your attack becoming shorter with velocity as well. You'll have to deal with ENVCTL and its attack fields. All in all, if u want to emulate realistically, you'll probably have to use several layers or even delve in the SETUP department.
More power to ya.
Yash.
FILTERING
Date: Sun, 10 Dec 1995
From: Jim Michmerhuizen
Subject: filtering
On Fri, 8 Dec 1995, Larry Hall wrote:
>>>I think I may have overlooked something, but how would I go about
>>>"stacking" a low pass _and_ a high pass filter onto the same one
>>>layer program?
>>Use a bandpass. A lo and hi in series effectively ARE a bandpass
>>filter.
>well, this is true if the corner frequency of the highpass is LOWER
>than the corner frequency of the lowpass. for the opposite situation,
>you have something a bit different...
Well, yes. If the two filters are in series, you have an attenuator. If they're in parallel, you have a notch. Roughly. Somebody else pointed out earlier that there's some _practical_ difference, within K2k's architecture of "algorithms", between those that support or supply an explicitly bandpass function and those that supply a lo and a hi in series.
"Attenuator" at the beginning of the preceding paragraph, is sort of a joke, guys. It's like when you've got a single-pole lopass at say 20Hz. Lowering the cutoff another octave could only change the amplitude, not the timbre, of what's passing through. By the same token, if you've got a series setup as described (hi > lo), you'll just attenuate the signal, not change its spectral balance.
Regards
Jim Michmerhuizen
GATING ONE
Date: Thu, 30 Nov 1995
From: bennitt_j
Subject: Gating effects....
I had this problem on my K2500R, the A Clock, BClock etc do synch to the MIDI clock sent from my computer but there's no set point that the A Clock (etc) start from so if you wish to gate a drum loop or whatever it's useless. I've e-mailed Kurzweil directly and they've responded to all of my questions except one concerning the A Clock, I've been promised that they're looking into it.
In the meantime I have two methods of doing exactly what you're after,
1) I send the audio out to the audio IN of an analogue monosynth (Roland SH09) and gate this by triggering the Roland from a MIDI to CV convertor (this is cool as I have real-time tweakability over the SH09s 12dB filter and envelopes).
2) I use MIDI volume control commands to send 127 and 0 values (for ON and OFF equivalents), this is easy from Cubase (and I guess other computer based sequencers) but could be tiresome from some hardware sequencers that I've used. The cool thing about not using a function as limited as A Clock (etc) is that you're not stuck with straight 16ths for the gating effect, if you want you can shorten/lengthen/miss out particular notes for a more interesting effect. You can also cut & paste the volume control changes to other MIDI channels so that all of your MIDI sound generators follow the same pattern (mix this with MIDI panning controls and weep).
I'll let you know if Kurzweil get back to me regarding reseting the A Clocks so that they start at a defined point each time you start and stop the MIDI clock source.
J.Bennitt
GATING TWO
Date: Fri, 1 Dec 1995
From: Sabirsh Al
Subject: a long(ish) post about CLOCK controllers and gating
The first thing I reacted to was Zaque's statement about CLOCK controllers: "they don't seem too useful". Not so; they are very useful, but limited in function. I shall explain, and assume you have version 3.x.
The first thing to do is to do is to go to the MISC page under SONG, and adjust the SYNC parameter to INT (internal). If it is set to OFF, all clock controllers will be disabled. If it is set to EXT, then you need something hooked up to the K2K MIDI in that is sending MIDI clock information (most sequencers do this, so does the Lexicon Jam Man). So pick the relevent SYNC setting - I'll use INT because with EXT the sequencer has to be *running* - while its stopped most of them don't send MIDI clocks, which can be inconvenient for programming.
To do what you want to do Zaque (tempo based gating), I use the final AMP stage. Take programm 199. Go to the final AMP page. Set control source two (or whatever its called - far right second from the top) to say, CLOCKB and set it to on. Then set the maximum value to say 24dB (this might be good to try at low amplification levels). You should hear the sound beat or pulse. If you adjust the minimum value to -96dB, the sound will actually stop periodically. And voila: a gate. Like I said, this is by memory, so fiddle with the maximum and minimum values to get the result you want.
These controllers can also be applied to pitch and filter parameters to give wonderful synchronised appegio-like effects. I'll upload some if anybody is interested.
Zaque, you were correct about the FUNs. There is no way to use FUNs to quantise MIDI clocks, and there is no way to sync LFOs to MIDI clocks :(
Bummer, but there it is - there *are* some things the K2X can't do. As I mentioned, the CLOCK controllers are also limited. You can only cause changes that occur with every beat, or fractions of every beat (1/2, 1/4), but *not* every fourth beat for example. There are two work arounds, both involving sequencers, and none involving FUNs. The first is to time-compress all the midi events so that they only take 1/4 as much time, and then reduce the tempo of the song to 1/4 the original tempo. The result is the song sounds the same, but MIDI clock messages are sent out at one quarter the original tempo. The other way is to use the method of J.Bennit and send volume changes as controller values. This is really the best because you can match those controller messages to groove quantise functions and have gates that match your groove, rather than being rigidly quantised.
Neat eh?
However, if you want to use CLOCK controllers with a sequencer (by setting the SYNC function on SONG/MISC page to EXT) your gates, filters, pitches whatever will allow follow MIDI tempo changes. However, J.Bennit pointed out that these functions "have no set point that they start from".
What this means is that that controller functions go out of phase if you stop your sequencer in the middle of a song and then start it again. They are still synced, and will change in tempo, but the maximum value of say ACLOCK will no longer come at the end of every beat, but maybe in the middle, or at the beginning, or whatever. In my experience the work around for this is to always start the song from the beginning, because then everything will always be in sync. I suspect this has something to do with either how the sequencer is sending MIDI start/stop messages, or how the K2X is receiving them. (I use Cubase by the way - which may be important because of the resets that are transmitted when the sequencer is reset to the beginning of a song).
I hope you are less confused now than when we started.
AL
GAIN STRUCTURING
Date: Wed, 26 Apr 1995
From: David Fox
Subject: Gain Structuring
One of the software engineers has asked me to pass along this advice re: getting the best signal to noise ratio.
1) Reprogram the internal effects. In the 2500, the internal effects are much quieter, and sound better than in the 2000, even though the 2500 uses the exact same chip in the 2000. Most of this is accomplished by by turning up the effects' output levels. For example, Sweet Hall has it's Early Level and Later Level both set to 7. Turn 'em up to 10. (11 would be even better.) For some presets it's not quite so simple due to the required balance of reverb to chorus or whatever, but usually some improvement can be made.
Programs that use these effects will then need to be reprogrammed to use less. Reduce the wet/dry mix by ear until you obtain the old or desired balance this is quite a lot of work, but the results are really worth it.
2) Turn up the output levels of your programs to make them hotter. If you have version 3 you don't need to reprogram, just go to the MIDI:CHANNELS page and set each channel's OutGain to 6dB. This uniformly adds 6dB to every layer regardless of the gain structure. If you have version 2 or older, this parameter acts a bit differently, so you should go to the output page of each layer of each program and increase by one notch (add 6dB).
You can go even hotter. Try adding 12dB instead of 6. If you're not playing chords, you could go even hotter (18 - 24dB). When you are using a lot of voices, you will experience some clipping in certain situations, but in those situations, S/N is usually not a problem. Use MIDI volume to turn things back down.
GRANULAR SYNTHESIS
Date: Thu, 10 Aug 1995
From: Jennifer Hruska
Subject: Granular Synthesis
46rom the MIT Press: "Representations of Musical Signals"
Giovanni De Poli
"Granular synthesis is an innovated approach to the representation and generation of musical sounds. The basic idea is that a sound can be considered as a sequence, possibly with overlaps, of elementary acoustic elements called grains. Granular synthesis constructs complex and dynamic acoustic events starting from a large quantity of grains. The features of the grains and their temporal location determine the sound's timbre. We can see it as being similar to the cinema, where a rapid sequence of static images gives the impression of objects in movement."
..."In music, granular synthesis arised from the experiences of taped electronic music. In the beginning musicians had tools that did not allow a great variation of timbre, for example, fixed oscillators and filters. They obtained dynamic sounds by cutting tapes into short sections and putting them together again. The rapid alternation of acoustic elements gave a certain variety to the sounds. The source of the elements could be electronic or live recorded sounds that were sometimes electronically processed. Xenakis (1971) developed this method in the field of analog electronic music. Starting from Gabor's metoid, Xenakis considers the grains as being music quanta and suggest a method of composition that is based on the organization of the grains by means of screen sequences, which specify the frequency and amplitude parameters of the grains at discrete point in time."
..."Other methods of synthesis, based on particular forms of elementary waveforms, have been proposed in computer music, especially as a way of realizing subtractive synthesis. The most important are VOSIM by Kaegi and Tempelaars (1978) and formant-wave function method of Rodet (1980). These methods can also be considered as particular types of granular synthesis.
In this case the temporal position of the grains is directly related to the pitch of the sound, and their waveform determines the spectral envelope. It can be pointed out that even sound synthesis from time-frequency representations can be realized in a granular manner as shown in the chapter by Grossmann and Kronland-Martinet and Arfib in part I. Granular synthesis is in fact the other face of additive synthesis, and it allows the production of the same sounds in an alternative way." Curtis Roads
..."Musical possibilities have expanded far beyond the trinity of "melody, harmony, and rhythm" to the point that our composing universe is truly what Varese called the domain of "organized sound" In this vast domain of acoustical phenomena, (editors note: VAST indeed! ..ahem..) machine assistance is not only welcome, it is a necessity. This is especially true of such computationally intensive techniques as asynchronous granular synthesis (AGS), where the artist's brush becomes like an extremely refined spray jet fed by an array of colored paints. In AGS each dot in the spray is a sonic grain, and the tone color of the grain is determined by both its waveform and frequency."
..."One of the most interesting features of granular synthesis is that we can insert any waveform into a grain, including waveforms extracted from recorded (sampled) sounds."
The paper goes on to describing different methods of implementing granular synthesis. The definition of granular synthesis lends itself to many synthesis techniques by its very nature. Musicians have done it by splicing tape together in unusual sequences, computer musicians may do it by generating values for each grain and letting the computer crunch away at it. Granular synthesis and the K2000? My first thoughts are that it is possible to do this in the Sample DSP Editor. Using the MixBeat, Replicate, MixEcho, and BeatVolAdj functions you could select "grains" of sampled data and string them together at regular or irregular intervals, thus generating a sort of granular sequence. Although "grains" connotate a very small amount of data, perhaps a single cycle waveform, it wouldn't necessarily have to, although I suspect the smaller you go with your grains, the more your sound really synthesizes into some new. Some experimentation here could be very interesting.
-Jennifer Hruska
Date: Fri, 11 Aug 1995
From: Dariusz Garncarz
Subject: Granular Synthesis
hi Jennifer...
i'm the original granular synthesis & creating worlds postee...
i really enjoyed your explanation of granular synthesis...
i've done some work with granular (i sued it for simulating LONG whale calls)...and i've worked with people that use it constantly...
i've written a granular synth myself (for the PC), and used what i've learnt from barry truax (the granular artiste - prof)...
anyhow, getting back to the subject & that being the k2k:
the k2k can ALMOST do granular, expcept it AltSampStart switch should be turned to a coninuous control parameter (so one can offset into a sample)...then simply apply a 50ms ASR envelope and write a sequence (or MAX patch) that will trigger note on's every 50ms and you have a granular layer!!
i've written to david fox about adding this feature to the k2k (someday) and they're working on it....
if you like, e-mail mr fox and add your support of this feature (maybe then it'll get implemented into the k2k)!
cheers,
dAREk
Yes, we really need that AltSampStart to be continuous, I agree big time. I will make some calls, talk to some folks, ya know, and see if I can do (another) push for it. Could you elaborate on the method of granular synthesis you have done and how your PC program works? _jenn
Date: Fri, 11 Aug 1995
From: Dariusz Garncarz
Subject: Granular Synthesis
hi jennifer,
for my pc granular synth i used a regular synth model but with a sample offset and very short envelopes....i hardcoded the fact that when a note on is received, i start triggering an alternating seuqnece of short 50ms stabs (each playing for 50ms at 38ms intervals (for xfading))...you can vary the # of these granular layers, change the speed (0..110% of the original sample length), change directions (this is really cool... i can reverse a granular layer while the samples (grains) themselves are still playing in the forward direction)... this granular synth responds to the standard midi controller as well... it's kinda like barry truax's machine(s)...but they can sample something and start processing it all at teh same time...
cheers,
dAREk (Editors note: For more info on Granular Synthesis, Daniel Fisher, Head Sound Designer at Sweetwater, has an a reprint of his '98 Keyboard article at http://www.sweetwater.com/k2000/keyboard/ )
HIDDEN EFFECTS REVISITED
Date: Wed, 5 Jul 1995
From: David Fox
Subject: "hidden" effects
As stated by someone else on the list, the Digitech 256 comes with 127 preset effects. Since the Digitech has 127 slots, there are 127 possible effects that you can call up on the K2000. To spread those numbers out over the 10 banks, they are accessible using the following numbers.:
- 1-37
- 100-109
- 200-209
- etc
- up to 900-909
We overwrite the first 47 presets with our own effects settings (0-37 and 100-109). We create our own objects in the Setup ROM which cover up the ones Digitech created.
But the other 80 effects (#s 48-127) are still accessible by typing in the appropriate K2000 number (a number between 200 and 909). The K2000 wil say Not Found because there is no Kurzweil objct in ROM at that number. But it still sends a program change to the Digitech, calling up tha number. You must type the number in (ass opposed to scrolling with the wheel) because when scrolling, the operating system is designed to skip over "Not Found" object numbers. You could also send a program change (with FX Mode set to Master and sending the command on the FX Chan) to call up the effect.
Although I don't have a list of what those effects are, they are the stock presets that come with the Digitech 256, so you could find out by asking Digitech.
Keep in mind there is nothing "special" about these effects - they are simply the presets that Digitech chose.
INTERNAL EFFECTS (AGAIN)
Date: Fri, 28 Apr 1995
From: Da5id Din
Subject: Using the K2000's Internal Effects Processor...
Here's an article I wrote for KOG. The K2000 in the article is a keyboard version although this will work just fine with the rackmount. Enjoy!
Using the K2000's Internal Effects Processor as an Outboard Processor by David Spiegelman.
Have you ever wanted to use the K's effects processor on your other synths and/or drums machines? Or would you simply rather have more control over the effects levels of internal instruments? Well here's how.
To get started you will need a mixer with four available channels for your K (we'll call these inputs 1-4), one aux send, and a stereo aux return. In addition, you will need five 1/4" instrument cables and one stereo insert cable (I use Hosa's) with 1/4" connectors.
The hookup goes as follows. Using the insert cable, connect the stereo (TRS) end to output A (L), connect the send (tip/red) end to input 1 of your mixer and the receive (ring/grey) end to your mixer's aux send. Then connect MIX outs (L & R) to your mixer's stereo aux return. Lastly, connect A (R) and B (L & R) to mixer inputs 2 - 4 respectively.
All that remains is to check a few settings. First go to the MASTER page and set OutA->FX :L Only. Then go to the EFFECTS page and set Wet/Dry:100%Wet and FX Mode:Master.
What did we just accomplish? We now have four dry outputs from the K each with an adjustable effects level via the appropriate send knob. We also have the ability to apply effects to other instruments connected to your mixer. Have fun!
JUNGLE FEVER
Date: Tue, 21 Nov 1995
From: "HECTOR, KEVIN"
Subject: Jungle tips
OK. For a start, dont use that JB break with "You gotta" vocal in it, and don't use the Funky Drummer. Old and tired !
Try this. Get your break set up. Play the break through a delay unit (I think analogue sounds better) with tons of feedback and a very short delay time to get that almost flanged metallic sound (about 20mS should do for starters). Lay the FX returns off to DAT and sample this. Now change the delay time slightly up or down. It gives the metallic ring a different 'pitch'. Again, lay off to DAT and sample. Do this a few times to give a good variety of 'pitch'. Now splice up the new samples (using multiple sample objects pointing at the same sample data) and map over a keymap so that different keys start each 'pitched' break at a different hit (am I making sense here ?). With the original break spliced in a similar fashion on the same keymap (or a different one if you're going to process FX and clean versions differently). Now instead of your normal snare-roll fills you can almost play a melody with the break hits. With the right break (and a certain amount of restraint) some VERY effective fills can be built. Any break processing in this manner is fairly time consuming. You're best off doing this kind of work in one session then putting it aside as a toolkit for later use - it is not a particularily musical process.
This kind of multiple sampling of one break through various FX is quite commonplace in Jungle now and can produce some startling results. Another list member posted me with the following idea which sounds good :
>Take a breakbeat and feed it through an effect algorithm (like a
>large plate). Solo the effects and sample that. Now you have a
>sample of "drum loop ambience" which you can cut up and trigger at
>will while the main breakbeat (dry) is playing. I suggest sampling
>in stereo, and play the breakbeat in mono. When you trigger the
>ambience behind the breakbeat, the stereo field suddenly opens up –
>quite a cool effect.
>
> Shiuan
Although these techniques are not confined to the K2, it is very well suited to this kind of work, and the DSP functions - and I'm not just talking about filters here - add excellent dimension to your work and have a real "K2000 quality" signature. Listening to a lot of Jungle these days I'd say that the K2 is used extensively by many popular artists.
Kevin Hector
KAT Programming
Date: Wed, 6 Dec 1995
From: Joe Albano
Subject: k2000 and the KAT
>In glancing over the past several months of this list, I found no
>answers to the K2000's hi hat programability with the Drum KAT's
>"h.a.t." KAT trigger. Has anybody gotten deep enough (or obsessed
>enough) into this capability?
>If so, I would be thrilled to learn how to program the K2000 so
>that the KAT can realistically trigger 2 or 3 "Open" hi hat samples
>AND then shut off those "Open" hi hat samples with a "closed" hi hat
>trigger.
As you probably know, the HatKat pedal puts out continuous controller data (when used with an appropriate voltage-to-MIDI device like their DrumKat trigger interfaces, an Anatek voltage-to-MIDI box, or a (Lexicon) MRC). I used cc/1 (modwheel), but any cc# will do. (Don't use the switch output (middle) - make sure to use the continous controller output (closest to the front of the pedal.) In your K2k's drum program, make closed HH & open HH separate layers and assign them both to the same note(s). On the AMP page, use the following settings:
Closed HH:
Src1 : FUN3
Depth : -96dB (approximately, depends on the sample's amplitude)
Open HH:
Src1 : FUN3
Depth : 62dB (approximately, depends on the sample's amplitude)
For both layers:
Input a: Input b: Function: FUN3: =0.98 MWhheel warp2(a,b)
When the HatKat is pressed down, producing a value of 0, only the closed HH will sound; when it's opened up for a value of 1 (or 2) and higher, the open HH will sound. (This comes from the warp FUN - if you don't use it, the HH won't open till the pedal is 1/2-way up, unlike a real hat..)
Stomping down on the pedal cuts off any open HH layers sounding - once you get your timing down, it works like a regular hihat. A couple of caveats, though..
1) Don't grind your foot around too much when you're holding the pedal closed - this tends to send spurious MIDI controller data that make the HH layers sputter between open & closed
2) Be aware that each HH note you play consumes two voices of polyphony, as both layers are always active (it's just an amplitude crossfade) To have more variation in the open HH sound(s), there are a couple of options:
1) You could assign open HH decay to be controlled by cc/1, so opening the pedal would gradually increase decay time
2) If you wanted to use, say, a partly open HH sample (splash) and a wide open HH sample, make two separate open HH layers, one for each, and use the CrossFade feature on the OUTPUT page to fade between them. For example:
Partly open HH layer:
CrossFade : MWheel XFadeSense: Rvrs
Wide open HH layer:
CrossFade : MWheel XFadeSense: Norm
Keep in mind, now each HH note will eat up 3 layers of polyphony, since once again, all three layers are always active. To incorporate more variations of open HH samples, make more layers, and set up additional crossfades (or cross-switches, to save on polyphony - there's another trick you can use too, if you have the sampling option).
The final touch is to add another layer for the foot-closing sound, which would be triggered from the HatKat's trigger output (through the appropriate trigger-to-MIDI converter, of course). IMPORTANT - the open and closed layers must be assigned to the same MIDI note(s)! This is only a problem if you want to play this program from some sequence that doesn't use cc data, just the normal open/closed-HH's-on-different-notes approach (see below).
There are a couple of other tricks I use to tighten this up, and to make this work from a drum sequence that doesn't have any cc data, only different open/closed notes (or to have my cc-HH sequences play regular drum programs), but these tweaks utilize the powerful "Environment" real-time data-transform features in my sequencer of choice, Logic (Audio).
This approach has worked well for me, both from a keyboard and from a full kit of MIDI pads - maybe it'll work for you also.
Bye-
Joe Albano
PS-
I heard KAT just went out of business - too bad, they made good stuff..
KEY TRACKING
Date: Wed, 7 Jun 1995
From: Darwin Grosse
Subject: Question about KeyTrk...
>... However, I am a bit confused
>about key tracking and the way this relates to the coarse adjust for
>the lowpass filter. I'm just going to throw a bunch of questions at
>you because I think they are all related. If the keytrk is set to
>anything more than 0 cents, the cutoff frq. should rise as you play
>higher keys, right? How does this reconcile with the fact that the
>coarse adjust parameter appears to be fixed at, say, A-440? Either the
>cutoff frq is fixed, or it rises or falls depending on the amount of
>keytracking applied, right? If positive keytracking is applied, and
>the cutoff frq rises as you play higher keys, shouldn't the value of
>the coarse adjust shift upwards accordingly?
>
>Perhaps another way of asking this is: If I set the coarse adjust to
>A-440, and I set keytrk to a positive value (not necessarily 100
>cents), which key on the keyboard will reflect this cutoff frq? It
>shouldn't necessarily be the actual A below middle C, right?
>
>Anyhow, I hope you can see where the source of my confusion is. I think
>it has to do with the difference between the cutoff frq. being
>"fixed" and being "relative" to what key is being played, and how the
>"fixed" interacts with the "relative."
Basically, whenever coarse and fine parameters are available for a parameter, these are considered the "base starting point" for that parameter. Any other modulation to the parameter will be adjustments from that "base" starting point, and will adjust according to the modulation settings.
Also, realize that the KeyTrk modifier is simply a pre-defined modulation path. The way it works (and someone, please, correct me if I'm wrong) is that the keyboard is considered a controller, with a value range from 0 to 127. The "center" for the KeyTrk controller is C4, with notes above this key generating positive effect, and notes below this generating negative effect. If the Filter KeyTrk value is set to 100 ct/key, each key will adjust the "modulated filter cutoff value" by +/- 100 cts for every key above or below C4.
In the example above, a filter is set with coarse @ A-440hz. Hit an octave above C4 (i.e., C5) and the following will happen:
- The filter's base cutoff frequency (coarse and fine settings) are retrieved.
- The note value is determined, with a "key difference" value calculated.
- In this case, that value is +12 (12 keys greater).
- The KeyTrk amount is used to calculate the filter offset: 12 * 100 1200 cts.
- The new filter cutoff is calculated by adding the base cutoff freq, and the filter offset: A-440 + 1200 cts A-880 (one octave higher).
You will note that this is roughly the same process that any modulation goes through, with the exception that "number of keys different" is used, instead of the "percentage of range" mose controllers use.
This should make the issue either crystal clear or mud murky. I'm at work, without the manual - if anyone can correct any problems with the above, please post. I'm just working off the top of my head here...
[ddg]
Date: Wed, 7 Jun 1995
From: Larry Hall
Subject: Question about KeyTrk...
regarding darwin's explanation of the keytrk parameter:
looks like a functionally accurate explanation on KeyTrk as applied to filter corner frequency. It might be added that for negative key track values, the filter cutoff frequency gets lower as you move up the keyboard. for a keytrk value of 100 cts/key (as in the example) the cutoff frequency will exactly track the keyboard. for positive values less than 100 cts/key, the cutoff frequency will rise as higher keys are played, but won't fully track the keyboard. a very useful function, indeed, and for things other than cutoff frequency... i use this parameter a lot in the shaper block for varying the shaper's adjust parameter across the keyboard, usually with a very small negative value to decrease the harshness of the shaper's impact on the upper registers of a sound.
--larry
LAYER CUTTING
Date: Thu, 18 May 1995
From: Christopher List
Subject: Getting one layer to stop another layer's sound (was something else)
"How can I get one layer to stop the sound of another layer, like a closed hi-hat stopping an open hi-hat?"
This is something that bothered me about the K2k keymap structure as well, here's what I came up with as a sort of work around...
- Set up a layer with a long release time from low C to D#4 and on the amp. page set a control source of FUN2 with a mod amount of -96db.
- Now go to the FUN page and set up FUN2 with: aGKEYNUM, bON, function aANDb.
- Now when you hit a note (don't hold it down) in the layer's range (say B3), then, while it's in it's release hit any note above E4, hitting the note above E4 will mute the other note (you could set up a different layer from E4 - HiC with another sound if you like).
Obviously, this isn't perfect, but it's a start.
Note that the GKEYNUM seems to get mapped to a value between 00 and 1.00 when used in a FUN, but the manual says it maps between 0 and 128. Obviously E4 .5 or .51, 'cause that's when the aANDb becomes active. This makes it hard to track one specific key as the first cut off key.
Unfortunately there's no FUN called "ab", cause then you could define one note to activate the function (there's soo many cryptic functions and so few simple ones...) - with something like aGKEYNUM and b.45 or something like that. By combining several functions, you could get something like this (ab) but I've been too lazy to figure it out.
If you set FUN2: aFUN1, bON, Func aANDb then set FUN1: aGKEYNUM, b.2 (or whatever) and Funca-b or a-b/2, you can change the range of keys that cause the volume to shut off. Using this system, though, it will always be everykey ABOVE some threshhold.
Note also that with this system, if you hit B3, hit E4 to mute the sound, then hit A3, the volume will come back up for whatever is left of B3's release. -If anybody wants to figure out a group of functions such that the last function evaluates as follows: if ab return 1, else return 0
I'd be glad to use it :) - I'm too lazy and too busy right now.....
-Topher
LCD INFO
Date: Thu, 18 May 1995
From: David Fox
Subject: K2000 Display issues
Here is some info about the K2000 display backlight and backlights in general. First, the K2000 uses what is known as an "Electroliminescent backlight" (EL). This type of display is very commonly used and found on many products. This type of display is characterized by extreme thinness (less than 1/16 inch), low power consumption (no perceptible heat generation), uniform brightness across the entire display, and relatively low cost. They are also limited in colors (basically a pastel green or blue). The issue with this type of display technology is that over time, the display will fade in contrast.
Thus the key to a long useful life is to keep the power off when not in use! However, even if you have the unit on 8 hours a day, seven days a week, you should still get many years of use out of the display before the contrast starts to fade. If you are having problems with the display and have not had it that long, it may be that it needs to be adjusted (regulating the voltage that goes to the display). This is something that should only be done by an authorized service center.
Other types of displays can not be substituted for the display in the K2000.
Other display technologies are thicker and there is simply no room for them in the K2000. Plus some consume more power than the K2000 can handle.
The K2500 uses a different type of display - a flourescent tube display.
Fluorescent tubes (CCFT) offer the highest brightness, a wide variety of colors, moderate power consumption, and relatively long life. However, they are a much greater thickness than EL displays and require very high operating voltage (approaching 1,000 volts!). They are also much more costly. Fortunately for us, heavy usage in notebook computers has recently increased the availability and lowered the cost enough for practical use of a CCFT backlit display in the K2500.
As stated above, the K2500 type of display can't be used in a K2000 or K2000R. The substantially greater thickness of the display module means it simply won't fit. Plus there's that 1,000 volt power supply to contend with.....
PANNING
Date: Wed, 15 Nov 1995
From: Larry Hall
Subject: Panning
>Could someone give me detailed advice about how to pan a single loop
>triggered once, through multiple outputs(at least 5) of the K2000 RS.
>I need to manually control the movement of the sounds through a
>surround sound rig.
- first create a keymap which contains the loop you want to pan 6 directions.
- assign the output to the A outs.
- duplicate the same keymap twice, assigning the outouts to the B outs for one and the C outputs for the other.
- now create a program with 3 layers, assigning each of the keymaps you just created to one of the program layers. now assign an algorithm containing the panner to each layer, and assign a different controller to the pan control on each layer and the output level on each level.
- an alternative would be to assign an algorithm with the upper/lower amp.
with this approach, you would assign a separate level control to each of the 6 upper/lower amps. i would personally prefer this method, as it would give me discrete control of each of the 6 outputs' levels, as opposed to having 3 sets of volume/pan pairs. it's 6 controllers any way you slice it... hope this helps - kind of sketchy, but should be enough to getcha going...
--larry
PITCH – ENDLESS SCALE
Date: Wed, 31 May 1995
From: Bill Simpson
Subject: K2K programming -- endless scales
Endless Scales
A few years ago, I heard a piece of electronic music which created the illusion of continuously falling pitch for an extended period of time, but actually the pitch ended up at the same level it started. The effect was quite striking. It turns out that a similar effect can be created on the K2000, although the method so far works best for scales rather than continuous pitch changes. The idea is to create a series of partials at octave intervals, so that the pitch class is clear but the pitch height (octave) is ambiguous, and so that notes an octave apart have the same spectrum. The overall spectrum looks like:
|
| | |
| | |
| | | | |
. | | | | | .
| | | | | | |
. | | | | | | | | | .
C-1 C0 C1 C2 C3 C4 C5 C6 C7 C8 C9
Different pitch classes have different partials (e.g. D-1 D0 D1 instead of C-1 C0 C1 etc.) but the same envelope. The difference between the highest and lowest amplitudes of the envelope is large enough so that the extreme partials (e.g. C8 and C9) are actually inaudible, so that partials C1 through C9 would sound the same as C0 through C8 or C-1 through C7. On the K2000:
Starting with the default program, change the Keymap to SINE WAVE (163), leave algorithm 1, but change the DSP to Parametric EQ. Set PAD to 18 dB, center FRQ to C5 (523 Hz), width to about 2 octaves, and AMP adjust to 48 dB. (If the signal clips, reduce AMP but leave as high as possible, or edit the keymap and use volume adjust to reduce the volume the same amount for each sample root).
Set velocity tracking on the AMP DSP to 0.
Duplicate the layer two more times, and change KEYMAP transpose on one layer to 12ST, the other to -12ST.
Duplicate the whole 3 layer program two more times. Change one of the programs so the layers have KEYMAP transpose of 24ST, 36ST, and 48ST. Change the other one so the layers have -24ST, -36ST, and – 48ST.
Create an effect using Parametric EQ with Band 1 at .10KHz level –12dB, Band 2 at .50 KHz level +12dB, Band 3 at 2.26KHz level -12dB.
Create a setup using the 3 sine wave programs, the Parametric EQ effect, and 100% wet.
Use anything else in your audio path (EQ, tone controls) to cut the high and low frequencies.
On the K2000, there should now be a range between about C4 and C6 where notes an octave apart sound about the same. (It may also help to detune the layers using PITCH Fine -- sometimes layers with simple waveforms at exact multiples of pitch seem to interfere and give weird sounds.)
If you keep playing an ascending/descending chromatic scale and skip down/up an octave at one of the notes that sound about the same an octave apart, it sounds like a continuously ascending/descending endless scale.
Try it, and see if you can come up with changes to enhance the effect.
Now for the challenge -- It should be possible to use the same principle for a continuous pitch glide. My initial attempts haven't been very convincing. The problems are
1) the continuous controllers on the K2000 don't have very fine resolution, so if you set the controller for a very wide pitch range (e.g. -7200 ct), you get discrete changes;
2) if you drop the pitch by many octaves, the high partials are no longer in the inaudible part of the envelope and you lose the effect. It should be possible to have each layer have more than one partial (using keymaps for partials 1 and 2, or 1, 2, and 4, or perhaps by sampling the K2000 output), in which case each layer could be 2 or more octaves apart, and the total range extended.
Bill Simpson
SAMPLE START HACK
Date: Fri, 5 May 1995
From: Jennifer Hruska
Subject: Modulating SampStart hack
Since the K2000 doesn't directly support modulating sample start points, here is a way to sort of sidestep the problem. It's not ideal and it's a bit of work but it comes in handy for me sometimes. Works best when you've stretched some samples down low and the attacks get really sluggish but you still want to use the sound. Also good for creating sounds where you want to tie in velocity to start point.
Method 1: Create additional "soundblocks" (sample parameters in the EditSample pages such as Start, Alt, Loop, End points, Decay Rate, Release Rate, etc.) If you don't do any DSP you won't have to save the samples again, saving space. (works for ROM sounds too this way) First trim up your attack on a sample in the EditSamp TRIM page, save it to a new ID (saying NO to the save samples ?), and use that in a new keymap. Then you can do it again if you want, trimming a little further, save it to a new sample ID and keymap ID. Then in your program use velocity switching (or another controller) to switch between these new keymap layers. In effect you would have a sort of quantized version of modulating the sample start point. Along these lines you can also tweek the EditSamp MISC parameters:
Release Rate and Decay Rate, and then use the program parameters in EnvCntrl to multiply against those values. You probably want to keep your AmpEnv on "natural". (Remember EnvCntrl effects ENV2 and ENV3 also.)
Remember also that EnvCntrl won't make an effect on the attack (sinse it's multiplying against 0 in the "natural" envelope) but you're doing this with the sampstart points anyway. Does this make sense?
You are esentially using the SampleEdit parameters to define an ADR where the attack is really set by your TRIM Start of sample and the decay and release is set in EditSamp MISC page. Also, by setting the "A:" start point in the TRIM page, you can get two alt starts for the price of one soundblock, what a deal... In this case you could have two layers using the same keymap ID with one of the layers set to Alt ON.
Method 2: If you are wanting to tweek the sample AFTER the initial start time, like you would with AtkSegs 2 and 3, you could, theoretically, use the SampeEdit DSP functions to change the shape of your waveform. I know, a little crazy and quite time consuming but I've done it when I've been really desperate. Also, you don't have the DSP editor with ROM samples so this only works with RAM samples. All of method 1 would still apply.
A bit of work, a bit of a hack, but it works with your K2000!
-Jennifer Hruska
SAMPLING YER OWN OUTPUTS
Date: Tue, 19 Dec 1995
From: nursingh@CLYDE.ENS.INSA-RENNES.FR
Subject: Sampling its own output
To preserve your attack, you can insert a delay on the Layer level.
Go to the LAYER page or pages depending on how many layers your program contains , turn delay on and choose a min delay (say 3 seconds)
After sampling, you can edit your sample to get rid of the possible silence in front of the attack.
Yash.
Date: Tue, 19 Dec 1995
From: David Fox
Subject: Sampling its own output
The K2000 can sample its own output. However, there is a limitation. On the K2000, when sampling begins, all MIDI and VAST processing stops. Therefore, in order to sample its own output, you have to strike the key before you press record. This means that you will lose the initial attack transients. Once it begins sampling, it will play through the sample (and loop if it is set to do that), but any realtim VAST changes such as LFOs, envelopes, etc. will not happen. If you are resampling from RAM samples, then one way to avoid losing the initial attack transients is to use the Insert 0 DSP to insert a couple of seconds of silence at the beginning of the sample. Then you can strike the key and hit record, and still get the entire smaple. Of course, this will not work for ROM samples, since you don't have access to the DSP functions for ROM samples.
All in all, the best solution is to record the sounds to tape, then sample them back in. If you have a DAT, then you will not lose any signal quality. The K2500 does not have this limitation. Its Sample-While-Play feature allows you to play anything, even a sequence and sample it.
| Main Page : | Documentation : | FAQ : | DSP Blocks : | KDFX : | Algorithms |
