Kurzweil:Dead Letters 1995
From Sonikmatter
| Main Page : | Documentation : | FAQ : | DSP Blocks : | KDFX : | Algorithms |
1995 Part 1
Pulled for the Kurzweil List Archives by Brian Cowell.
Contents |
For Whom the Bells Toll
Date: Mon, 25 Dec 1995
From: Lynn Cooley
Subject: For whom the bell tolls (original, ain't it?)
Here's a small programming example. Dennis Walker had asked about a tolling bell sample (complete with both ding and dong). I opened my big mouth to say that maybe you could tweak one of the K2000 presets to do this. How?, says he, so i gave it a try. The result is not exactly carrillon-quality, but it was a fun exercise, and i'm passing it on for those who are interested. Don't say nothing happens on this list but flames. (And this isn't a fire bell, either.) This is from memory, so it's not exact. Feel free to adjust to taste:
- In Program mode, find the preset called Church Bells.
- Press Edit, and use the More buttons to find the DupLayer button, and press it. Layer 1 will be the ding, and layer 2 will be the dong.
- On layer 2, we need to delay the attack, to account for the time it takes the clapper to bounce off one side of the bell and swing over to strike the other side. Change attack segment 1 to 1sec, 0%. This will cause its volume to be 0 for the first second after a key is pressed. Change attack segment 2 to 0.02sec, 100%. The .02 is the smallest time you can dial up; if we leave the ramp time at 0, the second attack segment is entirely disabled, so we choose the smallest nonzero value instead.
- Use the More buttons to find the ENVCTL command, and go to that page. Choose a value like 0.9 in the upper left entry, which is the attack segment key tracking. The result of this is to increase the 1sec attack delay on the lower notes, and to decrease the delay on the higher notes. This simulates the shorter swing of the clapper in smaller bells.
- Use the More buttons to find the Keymap button, and go to the Keymap page (still in layer 2). Select the Timbre Shift entry, and dial up some number there . . . maybe i put in -30 for the heck of it. The purpose of this was to make a difference in the sound between one "side" of the bell (layer 1) and the other "side" (layer 2). It didn't make as much difference as i'd hoped, but there are more tweaks to come.
- Find the Pitch button, and go to the Pitch page (layer 2). Change the Fine Hz to -27 or so. This makes for more difference between the sides" of the bell, and also simulates a bit of Doppler Effect, where the bell swinging toward you sounds slitely higher in pitch, and the bell swinging away from you sounds slitely lower in pitch.
- Find the F2 FREQ page (i think that's the name of it; it should be the 3rd button on the page that shows F4 AMP). We want the "dong" to sound a little muffled, because the clapper won't strike quite as hard on the rebound. Adjust the frequency downward. I don't remember where it ended up -- maybe C3 -- but you can find a value experimentally: Turn the dial counterclockwise while you keep playing a note, until the sound drops drastically in volume (that's too far). Then keep playing and turn clockwise until the sound gets back to a more normal volume, but not as brite. Somewhere in there should do the trick.
- Next (also on the filter page, layer 2), select the velocity tracking parameter. I set that about 90 on layer 2, so that when the "bell" is struck hard, you can hear more high frequency partials.
- Switch to layer 1 (still on the filter page), and set the velocity tracking to . . . oh, maybe half that; i think i did 47. The "ding" side of the bell now responds well to velocity also.
- Finally, back to layer 2, find the Layer page and set the IgnoreRelease parameter to ON. This allows you to trigger a note and release it immediately, but the second layer will still play after the 1sec delay. This is an optional step. If you leave this parameter OFF, you can hold a key down to play both the ding and the dong, or you can release the key immediately to play only the ding. There. It almost takes longer to describe it than to do it. You can toll a few bells at once, but keep in mind that with only 24 voices in a K2000, you'll soon run out.
MERRY CHRISTMAS
Lynn Cooley
Polyphony Priority
Date: Thu, 30 Nov 1995
From: Joe Albano
Subject: Polyphony Priority Feature II (long post)
Re: Polyphony Priority Feature:
>> On my Roland XP50 you can decide which sounds will drop out last, when maximum polyphony is reached. Is there such a feature on the K2k?
OK, I looked over the old programs - here's what I could glean from the settings in them.. (I'm NOT at all certain about any of this, without getting the lowdown on the voice-stealing specifics from Kurzweil tech support, but..)
The two problems I was trying to solve back then were:
- A 4-sec crash sample that got cut off too easily as it rang out, and..
- A 6-sec ride sample that stole notes from other sounds too easily when I played repeating 8th or 16th notes on it
A few observations..
- Notes held down by keeping the key depressed have priority over notes sustained by the pedal (this is generally good - it means if you do a big flourish on, say, a piano patch with the pedal down, the earlier piano notes will be stolen rather than older sustaining notes on another track, which would be more obvious and problematic.
- For drums/percussion sounds, there's two triggering options (as usual):
Another option for drum sounds is gate mode, but with a release long enough to play the entire sample, even after the noteoff. This will seem to play just like trigger mode, but I think the priority of those notes with this method is different than with a standard trigger mode. (again, here I can't be sure of the specifics w/o more info from Kurzweil). I noticed my multi-layer drum programs mix these two approaches up, which I think at the time I had figured this stuff out, was done to manipulate the note-stealing priorites.
- As a note rings out (I think) it's priority changes as it goes thru the various envelope stages (ie, it has higher priority during the attack stages, lower during decay, still lower during release (just to be clear, I mean higher = less likely to be stolen, lower = more likely) This is like a more sophisticated version of the standard "last-note" priority scheme usually implemented, which I'm sure is also operating here.
Here's how I was able to make this information solve the above two problems:
- I put the crash sample layer into standard Trigger-mode - LAYER page, IgnRel: On ( IgnRel = Ignore Release). rather than using gate mode with a long release - the audible result was the same, but it didn't get stolen as easily (I think).
- I did the same thing with the 6-sec ride, BUT when I played quick repeating 8th/16th notes against, say, a piano track, the piano notes were stolen way too noticeably. This was because the build-up of long, overlapping rides quickly ate up all the polyphony, the voice-stealing algorithm kicked in, and the trigger-mode rides apparently had higher priority over the gate-mode, pedal-held piano notes. Even simulating an acoustic drumkit, I still didn't need 20 or so 6sec ride samples overlapping each other (3 or 4 would have sufficed), but putting the ride back to gate-mode/long release gave it too low a priority and then the rides were stolen too obviously.
The answer was to use the other two trigger-mode options on the LAYER page:
- ThrAtt or TilDec. ThrAtt (thru Attack) acts like IgnRel ONLY during the first of the three attack segments of the AMP envelope; TilDec (till Decay) acts like IgnRel during all three Attack stages.
So, if I used IgnRel, and set a (user) AMP ENV (1) to run at 100% level for 6 seconds (the full length of the ride sample), that note would have a high priority. If I turned IgnRel off, and set a 100% 6-second release time, that note would have a low priority. To get something in between, I set TilDec On; in the AMP ENV (1), I could then incrementally raise or lower the priority by balancing the length of the Attack vs. the Decay segments.
For example:
- 100% 6sec envelope with:
The note's duration is the same in all the examples of course, but you're controlling how soon it's available to be stolen and how much of the note will most likely play before it's stolen. I ran a test sequence with a typical arrangement including piano and these drums/cymbals, and adjusted this by ear until the note-stealing was working the way I wanted it to.
It's worked well for me 99% of the time since then.
Whew! That was pretty wordy - I guess it's no easy substitute for a nice straightforward "layer-Priority" setting, but at least there is some control available, if your programs are adaptable to it. In practice, Kurzweil's voice-allocation system works extremely well - only when I use lots of long, overlapping custom samples do I ever notice it working.
I hope this is useful..
Joe Albano
PS- another quick tip regarding polyphony in multi-layer (drum) programs..
Let's say you have a drum program with a cymbal layer - the keymap is like so:
C1 (lo) crash cym 1; C#1-B4 blank (silence or sample not found); C5 (hi) crash cym 2; LAYER range C1-C5
Another drum layer has toms in between from C3 to C4.. When you hit a Tom note, you use up TWO voices of polyphony - one for the tom and one from the overlapping "blank" (supposedly unassigned) cymbal layer.
Solution: always break up blank sections like this by making two layers (ie) cymbal 1 layer range C1-C1 only; cymbal 2 layer range C4-C4 only.
You can still use the same keymap for both, but since no layer range includes the "blank" area in between, no polyphony is wasted! (in certain cases, you might also accomplish this by utilizing the " Opaque Layer" feature)
Copy Protect Your Samples and Other Secret Button Presses Reveiled
Date: Fri, 8 Dec 1995
From: Ken Shvetz
Subject: Secret K2000 button presses!!
Hello, K2000 users!
I thought you might be interested to know that i've discovered what i believe to be two "secret" double-button presses. I didn't come across these anywhere in my manual, even after intentionally searching. If they ARE in the manual and i'm just a blind idiot, please flame me for not reading the whole thing front to back like most users do!! :-) Also, if they ARE in the manual, please tell me WHERE (!?).
Note that these tips apply to version 3.01J (or greater, i assume) and probably won't work on all earlier versions of the OS. Some further comments and questions appear below, but first, here are the two tips:
1. Copy protect your own samples!
On the Edit: RAMSample page (either MISC, TRIM, or LOOP), press the up and down Channel/Bank buttons simultaneously. The K2000 displays:
"Copy protect this sample?" ["Yes/No" soft buttons]
"Are you sure?" ["Yes/No" soft buttons]
"This sample will save as copy protected." ["OK" soft button]
(See below for more on this.) 2. Hard format a SCSI drive
On the main Disk Mode page (accessed by pressing the " Disk" mode button) likewise pressing the up and down Channel/Bank buttons simultaneously will display an undocumented feature. It will only work if you have one of the SCSI devices (0-7) selected; pressing these buttons simultaneously does nothing if " Floppy" is selected.
The K2000 will ask:
"Hard format this device? It will take a while..." ["Yes/No" soft buttons]
Now, BEFORE you run out and try this, let me make a suggestion: Set the drive to a SCSI ID to which nothing is hooked up. That is, don't do it with your HD selected! I can't be responsible if you accidentally press "Yes" and then erase your whole hard drive!
Further comments...
Regarding Tip #1 (and un-copy protecting samples!):
According to the K2000 manual, copy protected samples (from other manufacturers) will not Dump. I don't know if copy protecting does anything else to them, but this is correct. You'll notice that if you now press Dump, nothing happens. This isn't any huge incumberance, since any sucker can still load them from disk.
However, they can't be dumped to other samplers. At least you can feel cool about having your own "copyrighted" samples! :-)
Furthermore, the discovery of this option a while ago led me to wonder:
If there's a way to copy protect samples, is there a secret way to UN-copy protect them!?!?! (Performing the key press again over a prev-iously copy protected sample won't display the question.) I haven't discovered any button presses for this. However, there IS a way that i think you can "un-copy protect" samples! This should work if you come across any samples that are copy protected (maybe Akai, Roland, Ensoniq imports that won't dump?):
&rarr To un-copy protect a sample, just export it as a .WAV or .AIFF file!
Then, reload it into your K2k and save it normally. Or, throw it on the Net and get sued. (Just kidding!)
Regarding Tip #2:
I have not answered "Yes" to any hooked up device yet (i don't know about you, but i don't have any empty HDs lying around!), so i don't know what (if any) messages follow. I also don't know how this differs from just pressing the " Format" soft button on the Disk Mode page, as i don't really know what a "hard" format is (as opposed to a soft format, i assume)! I've heard of a low-level format, but not hard and soft ones. Maybe it has something to do with overriding HD firmware or something. For example, my HD has its own error- correction scheme. The manual states that this is completely "transparent to the user," meaning that if disk errors occur, the drive will take care of it without the data getting messed up. Maybe if you hard format the drive, error correction would then be the responsibility of the K2000, for example?? I don't know. Someone enlighten me.
Ken Shvetz
TB303 Emulation 1
Date: Sun, 19 Nov 1995
From: glenn gutierrez
Subject: ESQ'1 <> TB303
from the eps user's group list:
>>>the ESQ has a small wavetable of 8-bit samples, an 8-voice (3 osc
>>>per voice) archetecture, and 8 nifty little Real analog curtis
>>>filters (for a warm sound).
>>My friend and I used to use an ESQ and a Mirage for our music (if
>>you could call it that 8 years ago). True the bottom end is not
>>full at all, but....did you realize that Curtis filters are the
>>same (by name, anyway) as found in the legendary TB-303 from
>>Roland? I wonder if the ESQ could do a convincing simulation.
>I'm not sure about the ESQ doing a 303 emulation due to the Accent
>or Emphasis section of the 303's VCF ( I'm not really interested in
>the 303 cause I'm not a D.J.)
>P.S.
>Are you sure the 303 uses a curtis chip as a VCF (I think it is a
>VCO like >in the following models)
>
Model VCO VCF VCA?
MC-202 1 CEM3340(x1) IR3109 DISCRETE BA662
SH-101 1 CEM3340(x1) IR3109 DISCRETE BA662
I read on the alt.music.makers.synth newsgroup about a study that some university had commisioned, looking in to the 'effects' of the TB-303 on the subconscious mind (!!!). Whatever the outcome, it did describe the innards of the thing saying that it used a Curtis filter chip - and that it *was* unusual to do at the time. If it's wrong, then it is. How'd the first guy on this thread get his info, I wonder.
Now, I'm not going to get into the many discussions possible:
- Is the 303 musically useful? (depends on your song)
- Is a "DJ" the same as a "dance music producer?" (one aspires to be the other)
- Doesn't the 303 sound like a juicy, wet, tuned fart? (isn't it great?)Aren't these all just subjective generalizations? ...
>From what I remember of the ESQ'1, the routing was fairly extensive.
In fact no "synth" since has really satisfied me until the K2000. I also used to own a TB303, until it was stolen. But from memory ...Emphasis Resonance. Accent = VCF Env Amount or VCF Cutoff, but probably the latter, thus preserving the full dynamic of the envelope. Sequencing all of the above is just as important as the sound itself.
I've just written a near perfect sim on the K2000 - I couldn't believe how bad the ones out there were. Velocity controls the filter cutoff "Accent" while the Mod Wheel controls the "Emphasis." Plus, there are a few little subtle secrets that really bring it to life. Hint: The 303's VCF envelope slope is not linear and the resonant frequency is not right at the cutoff frequency - that's too clean. Of course, this sim is of one type of TB-303 bass patch, hence the name PLNET303.
I was just wondering if the good old box from the past was capable of the things that technology had to come this far to finaly realize. Hmmmm....that'd be funny!
glenn gutierrez,
TB303 Part 2
Date: Wed, 26 Apr 1995
From: Jon Drukman
Subject: a lesson in 303 emulation
for those techno boffins out there in search of the ultimate Roland TB-303 simulation, i had a major revelation last night and i'm here to share it with you...
understanding the capabilities of the 303 is critical to convincingly faking it.
for example, the 303 has only ONE envelope which simultaneously controls filter & amplitude. if you cobble this together on the k2000, the results can be startlingly convincing, particularly if you modulate the envelope.
here's the basic idea:
- A 303 has one oscillator that can generate sawtooth or square, so start with default program 199, and assign Sawtooth or Square on the Keymap page.
- Choose the Four-pole Lowpass with separation algorithm.
- Go to the AMPENV page and choose User. set Att2 to something like 0.50.
- Set Dec1 to 0%. this step is very important; it gives you a sound that sounds for a fixed duration no matter how long you hold down the key. (for those used to traditional Attack/ Decay/ Sustain/ Release envelopes}, Att2 is the Decay.)
- Go to the ENVCTL page and set up some controller to modulate the attack. this will give you the "envelope amount" parameter. this controller will be used to lengthen and shorten the duration of the note. it will also have the effect of changing the filter sweep amount (when we get around to setting that up in a few moments) which is critical to getting some of the 303's distinctive sound.
- Go to the filter cutoff page (F1 FRQ) and select some low value for the base amount (i use around 200 Hz). for mod src 1, select AMPENV and something like 5000 ct. adjust to taste. you can also route another controller to mod src 2 if you want to have control over the filter (and who doesn't?!).
- Set the filter res (F2 RES) to whatever you like. you'll probably also want a continuous controller for modifying this. most of the 303 effects you hear on record are combinations of filter cutoff & resonance adjustment in real time. remember that the 303 has *five* knobs that you can play with while it blips away... we could simulate all of them but i've found that filter, resonance and envelope decay are the most important.
- One of the other 303 features is called "accent"... basically a note can be normal or accented. when a note is accented, it's slightly louder and it opens the filter a little more. use the VelTrk parameter on the F1 FRQ and F4 AMP pages to fake this. make sure you program your sequencer to only send two different velocities.
- Finally, to get the all important glide... on the common page, set monophonic to on, portamento to on, and portamento rate to around 200.
- If you're playing from the keyboard you'll probably trigger some unwanted glides. sequencing the notes will give you precise control. another useful feature is the MIDI portamento controller (#65). turn it on when you want the glide, turn it off when you don't.
- On my setup, i had the mod wheel routed to filter cutoff, CS 1 routed to resonance and CS 2 routed to envctl. those of you with the k2000 keyboard version will have to work out some other scheme since you've only got one control slider.
Jon Drukman
Simms and Compatibility
Date: Fri, 29 Sep 1995
From: David Fox
Subject: 8/9 bit SIMMs
The following information explains SIMMs compatibility issues in the K2000 and K2500, particularly with regard to differences between Calvin and Janis versions of the K2000. To determine which version that you have, simply watch the display while powering up the K2000. The K2000 will display the version number. If there is a "J" after the number, it is Janis. If there is no letter after the number, it is Calvin. (Information on the K2500 is at the end of this article).
There are two basic types of SIMMs - Composite and Non-Composite (often referred to as "True" SIMMs). Composite SIMMs are those in which there are more than 8 ( or 9 in the case of parity SIMMs) chips. Composite SIMMs use PALs to make several smaller DRAMs behave like a larger DRAM. These extra chips can cause timing problems with the 2000 and as a result, many Composite SIMMs will not work correctly in the 2000.
ALL Non-Composite SIMMs will work in either Calvin or Janis K2000s, as long as they are 120ns or faster. The K2000 was designed for use with 8 bit SIMMs* We do not recommend use of 9 bit SIMMs in the K2000. Some users have reported using 9 bit SIMMs without problems, but since we did not design the K2000 to deal with the 9th bit, we can not guarantee that there might not be problems caused with 9 bit SIMMs.
Currently, Non-Composite SIMMs are made in 1, 4, and 16 meg sizes. 2 and 8, and 12 meg SIMMs are always Composite.
When the 2000 was first released back in late91/early 92, virtually all SIMMs available in sizes larger than 4M were composite SIMMs. Even when 16M Non-Composite SIMMs did become available, they were initially much more expensive than the Composite versions. For this reason, Kurzweil worked directly with a couple of memory companies to develop Composite SIMMs which would work correctly with the 2000. These companies have long been listed in many places as approved vendors for SIMMs larger than 4M. Currently these companies are Custom Services and Simple Technology. It should be noted that these companies also make Non- Composite SIMMs.
When the Janis version of the K2000 was released, the sample memory subsystem was designed with less slack, so that SIMMs that were marginally slower than spec (but worked in most Calvin systems) no longer worked.
Composite SIMMs fall into two categories. Those that are 16M (call them type A), and those that are 2 or 8M (call them type B). Older versions of 1 and 4M SIMMs may also be Composite (type A).
Type B are made in "flavors" specifically for either Janis or Calvin. Each flavor will only work in the appropriate version of the 2000.
The confusing part of this is that in K2000 pre version 3, the K2000 would mistakenly think that the type B's were working when they were not. The resulting problems would be subtle (and sometimes not so subtle) corruption of samples. With version 3, if the type B are the right flavor, they will be recognized by the 2000, if they are the wrong flavor, they won't. Some users were frustrated with this because they had the wrong flavor of type B SIMMs with V2, and they seemed to work. When they went to V3, the SIMM's "stopped" working. In fact, those SIMMs really were not working reliably in spite of appearances and need to be swapped for the right flavor.
For type A Composites, both approved vendors currently make SIMMs which will work in either a Janis or Calvin unit. SIMMs made for Calvin units before the release of Janis will not work in a Calvin unit. (Custom Services will convert their older Calvin SIMMs so that they will work in a Janis unit for $30.)
Due to the fact Non-Composite SIMMs are now readily available and priced similar to Composite SIMMs, Young Chang/Kurzweil no longer recommends the purchase of Composite SIMMs for the 2000. Composite SIMMs consume several times more power than non-composite SIMMs. A fully loaded K2000 with 64MB of Composite SIMMs consumes TWICE as much power as the same K2000 with the SIMMs unplugged. Composite SIMMs raise the lowest AC line voltage that the K2000 will operate properly by a few volts. This can cause the K2000 to reset itself if your line voltage is abnormally low or you are operating off of a poorly regulated generator or a long extension cord. Composite SIMMs increase the heat inside the unit which can cause problems when working in hot environments* The added heat can potentially reduce the life of compoents inside the K2000, although this affect is hard to quantify. If you are going to purchase Composite SIMs, we suggest that you do so only from the formerly mentioned approved vendors, who have created SIMMs that have been tested for use with the K2000.
K2500 SIMMs requirements
The K2500 will accept ONLY Non-Composite SIMMs, 80ns or faster. This speed requirement is faster than on the K2000. SIMMs for the 2500 can be either 8 bit or 9 bit. Since 2 8, & 12 meg SIMMs are always Composite SIMMs, they will not work in the 2500.
David Fox on SCSI Management
Date: Tue, 27 Jun 1995
From: David Fox
Subject: SCSI info
Comments: To: K2000 List <K2000@american.edu>
As several folks have stated, it is often best to keep your SCSI chain as simple as possible. SCSI is a less than perfect protocol and problems can easily pop up. But contrary to Bruno's experience, it IS possible to have a complex chain and ,make it work. I myself have a chain which contains all EIGHT devices and never have any problems. I have a chain which includes a Mac, the K2000, the K2500R, plus internal hard drives for each of those units, plus a couple external removable drives.
The most important thing seems to be the quality of the cables that you use, along with their length. Also, you have to be careful to follow the procedures for how to deal with the Mac. Following is a document we have created with SCSI info:
****** KURZWEIL K2000 SCSI HELP ******
The following document contains information on using SCSI with the 2000, as well as a specific sections dealing with the Mac and the 2000.
Here are some basic guidelines to follow when configuring a SCSI chain:
- According to the SCSI Specification, the maximum SCSI cable length is 6 meters (19.69'). Users should limit the total length of all SCSI cables connecting external SCSI devices to Kurzweil products to 17 feet (5.2 meters).
To calculate the total SCSI cable length one must add up the lengths of all SCSI cables, plus 8" for every external SCSI device connected.
- The first and last device in the chain must be terminated.
Poor termination is a common cause of SCSI problems. If one installs an internal drive that is terminated, the termination resistors should be removed from the K2000 Engine Board. Having more than two terminators on the bus will overload the bus drivers, but this should not cause permanent damage to the hardware. Poor termination will corrupt that data on one's disk, however, as can bad SCSI cables.
For the K2000R, if it is not located at one end of a SCSI chain all internal termination, including the terminator resistor network on the K2000 Engine Board plus terminator resistors in the internal SCSI drive must be removed. It is much simpler to just make sure that the [[kn1|K2000}} is at one end of the SCSI chain.
A note about active termination - The K2000 uses active termination of the SCSI bus. Active termination has some benefits over traditional passive termination. Some people have positioned active termination as a panacea for SCSI problems, but this is more hype than reality. Active terminators are available to be used at the end of one's SCSI chain and all APS SR2000 series external drives use internal active termination that can be switched on or off.
- Each device in the chain (including int. hard drives) must have its own unique SCSI ID. The default K2000 ID is #6. Macintoshes use ID #7 & 0.
- Use only true SCSI cables - high quality, twisted pair, shielded SCSI cable. Do not use RS432 or other non SCSI cables.
The majority of SCSI cables we've tested were poorly made and could cause damage data transferred to and from the disk. Nearly all the SCSI data problems Young Chang's engineering department has had have been due to bad cables that didn't twist pairs of wires properly.
Correctly made SCSI cables have one ground wire for every signal wire and twist them together in signal/ground pairs. Cables made by APS Technologies (800.233.7550) are very good and are highly recommended.
Good cables are essential to reliable data transfers to and from the disk drive.
- One should buy all SCSI cables from a single source to avoid impedance mismatch between cables.
- Theoretically all eight SCSI IDs can be used. However, feedback from users has shown us that many people have problems with more than 5-6 devices in a chain. If you have 7 or 8 devices and are having problems, the best bet is to make sure you have followed all of the previous information, especially with respect to cables.
- Connect all SCSI cables before turning on the power on any equipment connected by SCSI cables.
- Authorized service centers will remove termination from the K2000 when installing an internal drive, set its ID correctly, and terminate the drive.
- When using a Macintosh, power up the K2000 and other devices first.
- The K2000 file format is a proprietary format; no other device will be able to read or write a Kurzweil file.
- The floppy disk format of the K2000 is DOS. The SCSI disk format is a proprietary form that is close to DOS, but it is not DOS. However, if you have version 3, the K2000 can read and write to a DOS formatted disk provided it was formatted on the PC with no partitions.
- It is possible to view, copy, move, name, delete files on a K2000 formatted floppy disk or removable media hard drive, with a PC or Macintosh running a DOS mounting utility program such as Access PC.
- As long as the SCSI bus is properly terminated there is no way that a user can damage one's hardware simply by operating it. There are a few hazards K2000 users should be aware of, however:
The only damage that usually ocurrs to SCSI hardware comes from static electricity "zapping" SCSI connector pins when the cables are disconnected.
The silver colored shell of the SCSI connector on the end of the cable is connected to ground and is safe to touch, but the brass colored pins inside eventually lead to the SCSI interface chip and are vulnerable. One should discharge static from one's body before touching SCSI connectors by touching the 1/4" jacks on the rear of the K2000 or another grounded metal object. Any devices connected to the SCSI bus should be turned off when plugging or unplugging SCSI cables.
If the K2000 is connected to a Macintosh or PC one should make sure that the computer cannot access a SCSI disk at the same time the K2000 does (see below for more info on this). Those who occasionally want to share a drive, but don't want to take any risks would be best served by disconnecting and connecting devices as needed. If you want to share drive(s) often and cannot constantly disconnect and reconnect devices make sure the Mac or PC is really done with the disk before using the K2000. One should quit or exit from all running programs and disable screen savers, email, network file sharing, and any INITs or TSR's that run in the background. If the computer and K2000 access the disk at the same time there will be no damage to the hardware, but the bits on the disk, K2000, and computer memory can easily be corrupted. One may not know that damage has been done to these bits until weird things start to happen for apparently no reason.
The Mac and the K2000
- The Mac really does not like having another SCSI master on the bus (ie the K2000). It assumes that it owns the bus and it's drives, therefore it will not tolerate the situation where the K2000 is trying to talk to it's (the Mac's) disk. This suggests that you never want to select the ID of any drive mounted on the Mac's desktop. Even more fundamental is the problem that the Mac assumes that the bus is always free, so if it's OS tries to do anything via SCSI when the K2000 is doing anything via SCSI, the Mac will freak. The only solution is, wait until your Mac is completely idle before accessing SCSI from the K2000.
- The Mac and the K2000 cannot share a drive in any way, with or without partitions. If you are using a drive with removable media, you cannot easily switch back and forth between a Mac formatted volume and a K2000 formatted volume. To prevent problems, you will need to unmount the drive from the Mac desktop before switching to a K2000 format volume. The Mac will basically ignore the volume if it's not Mac format, but once you insert a Mac format volume, the Mac owns it. Don't forget about #1 above; inserting a cartridge will cause the Mac to access SCSI, so don't try to use the K2K at that moment.
- The only good reason for connecting the Mac and the K2K on the same SCSI bus is to use Alchemy or equivalent. If you're using a patch editor or librarian, you can just hook up via MIDI. Connecting via SCSI will allow fast sample transfers thru the SMDI protocol. In this type of configuration the easiest solution is to let the K2K have it's own drive, and the Mac have it's own drive.
However, after experimenting we have discovered that when using a K2000 with a Mac and a removable media drive in the middle of the chain, the following scenario will work:
Start with a Mac formatted cartridge in the drive. When you want to use the K2000, put the drive to sleep from the K2000. You can then change to a K2000 formatted cartridge and perform whatever disk operations you need. When you want to go back to the Mac, put the drive to sleep again, switch cartridges, and then wake up the disk by pressing Load. Of course the K2000 will tell you it can't read the cart, but the Mac will now access it fine.
Accessing a K2000 Internal Drive from the Mac
Access PC is one of the many programs for the Mac which allow it to format, read, and write to DOS floppy disks and removable SCSI cartridges. However, we have discovered that it is possible to format internal K2000 hard drives, even though the documentation claims to only support removable media (not a fixed drive). We tried this with v3.0, but other versions may also work. Because the program claims not to be able to do this, we do not necessarily recommend it.
The main things to remember is:
Never change the disk contents (i.e. save or delete files) from the K2000 when the disk is mounted by the Macintosh. If you do, this could easily lead to trashed files, directories, or even the entire disk. Access PC has no way of knowing when the K2000 has modified the disk structure, and it can just overwrite any state of the disk it thinks should be there. The safest thing is for users to only connect a drive to either the K2000 or the Mac, and not both at the same time. Of course, you can't always predict when a Mac will access its drive, and it doesn't do SCSI bus arbitration, so using the Mac while using the SCSI bus from the K2000 (i.e. doing a disk mode operation) is also a bad idea, and can cause the Mac to hang.
FX Processing
Date: Mon, 15 May 1995
From: nursingh@CLYDE.ENS.INSA-RENNES.FR
Subject: Please Kurzweil
>compressor
v2.xx and above provide the NORMALIZE function which does a similar job.
>EQ
You've got the FX EQ (v.noisy but efficient) and the ALG1 PARAMETRIC EQ DSP function.
>stereo image processor
There are PANNERS and separate controls for two outputs on some ALG's, plus the ability to playback stereo samples. If u r looking for drastic stereo imaging like the Roland RSS or Q-sound I don't think it can be done on the K2k.But then again, maybe I'm wrong.
> Phase correlation
What is it?
You can have a good dephasing effect on patches by
- using for example two oscillators with their pitches
- detuned like in
OSC1 : +12ct, OSC2 : -11ct.
nb : notice 12 AND 11, not +12 and -12. This enables for a slight beating effect, when crests and troughs of each wave coincide at common multiples of 12 and 11.( understanding of this part isn't essential for the effect to work )
- use an ALG with a NOTCH FILTER and use a slow TRIANGLE/SAW LFO as SRC on the CUTOFF FREQ. Similar effect as in 1 but different flavour.
We know what's to be tried when there's more than 1 way of doing something, don't we? Especially when the two solutions yield different sonic palettes.
YES, mix the two. Oh BLISS.
> limiter
Dunno.
If u boost your internal AMPS too much you'll get clipping.
Hahahahaahhahahahahahahahaahahahahaahahahahahahahahahahaha <evil grin>
Yash.
Date: Wed, 17 May 1995
From: Jason Alexander
Subject: Please Kurzweil
>> If u r looking for drastic stereo imaging like the Roland RSS or
>> Q-sound
>> I don't think it can be done on the K2k.But then again, maybe I'm wrong
Well, I haven't been able to get the K2000 to completely imitate an RSS or Surround Sound (tm) style of processing, but I can boast about two incredible effects that I programmed while designing DID soundbanks and more specifically a sample library of Wavestation sounds. I won't even try to describe it, other than it's purely magical sounding on etheral female vocals that contain long sustains. Running sweet voices like Mariah Carey and Enya through these two programs just sends shivers down every engineer's spine here at OutPOST. Infact, two of them commented on how not even the $xxxx Eventide can come close to delivering such a unique 'presence'. It's a delayed flanger with panning and tons of reverb, all user-controllable amounts of course. The delay is set in such a way that there are two taps back to back (seperated by 2 or 3 ms I do believe, this seems to provide the needed Surround Sound effect strangely enough) and the slight touch of flange does something totally incredible to female vocals, lemme tell ya! I drove my dance vocalist's voice through it for her current CD project, and we fell in love (well, I mean, with the music - I already did love her:-)
We've done some totally WICKED stuff with the effects and algorithms on the K2000 to make her vocals completely unique.
I know I sound like a broken record, talking about my Wavestation on a K2000 list, but may I share a production technique we used with you good folks?
We had this vocal percussion going on for a number she was doing that required some 'life' on top of a padded part - so we decided to muck about with some of the more 'dynamic' equipment in our arsenal. The Wavestation was top on the list for providing some cool sequences pumped through the on-board vocoder etc. If there are any WS owners out there, to give you an idea of what it was, we used a custom variation on the funky little number called 'Vel.DrumVocoder' (RAM3 - Patch #27) and programmed custom percussion wavesequences to 'drive' the vocoder's action on a very resonant sawtooth wave. Ok, so the fun begins with the help of the MAC sequencer, LOGIC AUDIO, where we slaved the Wavestations 'wavesequence sync' to the incoming MIDI clock. The same clock drove the K2000S, where we set, using FUNs, the playback of a sample of her vocal percussion to be 'limited' and 'gated' using combos of ENV 2, ENV 3 and FUNs driven by LFO settings using 3-step settings to completely mangle her voice, but in a very unique, quantized and syncopated way. Used the above mentioned '3d' styled effects program I am so proud of, and finally through it all into an outboard rack for fine tuning. The end result? You'll have to buy her CD when it's released:)
So, you say, THAT's what a crack-pot producer/engineer does for a living! heh heh!
Hope it inspires some of you, for I bet it left the rest of you mighty confused <grin>
Jason Alexander
Polyphonic Portamento
Date: Wed, 16 Aug 1995
From: Wim Verheyen
Subject: programming challenge, polyphonic portamento
Hello K2K'ers
I used to have a Prophet VS before the K2K. That thing could do polyphonic portamento. The K2K normally only allows monophonic portamento.
I wanted to do poly porta on the K2K.
There is an article in InGear (KOG's Newsletter), Fifth Gear of April 1, 1993 (seriously !) with an example by Dan Fisher. It works, provided you set Keytrk on the PITCH page to -100ct/key.
My solution is almost identical :
- Start from Default Program 199.
- Use keymap 152 Dull Sawtooth for experimenting with the pitch.
- On the PITCH page set Keytrk to -100ct/key, that disables keyboard tracking.
Src1 :FUN4 Depth :6500ct Src2 :FUN4 DptCtl:ON MinDpt:0ct MaxDpt:6700ct
Now some FUN programming :
FUN1: ON MWheel a-b
FUN2: 0.01 FUN1 a/2+b
FUN3: FUN2 BKeyNum lowpass(fa,b)
FUN4: FUN3 0.97 a*b
Explanation : BKeyNum is used for pitch information related to MIDI note number. There is some explanation of BKeyNum on page 20-23 in the Musician's Guide which IMHO is not correct. I think it generates a value of -0.5 in response to MIDI key number 0, a value of 0 in response to MIDI key number 60 (C4), and a value of +0.5 in response to MIDI key number 127, the key off value must be 0. (in fact, the values of -0.5 and 0.5 are just slightly higher)
That results in an effective range of just 1 for 128 MIDI key values, or 12800 cents for equal tempered tuning. That also means you would need a pitch modulation depth of 12800 cents.
Dan Fisher's trick is to apply 7200 cents to both Src1 and Src2, and scaling Src2 with DepthControl from FUN4, resulting in a modulation depth of 12816 cents.
My trick is to apply 6500 to Src1 plus 6700 to Src2 and multiplying the output of FUN3 (BKeyNum) by 0.97 in FUN4, that results in 12804, slightly better FUN1 inverts the action of the modulation wheel (Dan used the Data slider, that works equally well) from 0->1 to 1->0.
FUN2 applies a very slight positive offset so that the final 0 value is not reached, this is needed because otherwise the lowpass function takes forever with the modulation wheel all the way up.
FUN3 does the portamento : a lowpass or lag function.
Challenge : The pitch shift is from C4 upwards for notes higher than C4 and downwards for notes lower than C4. Is there a way to shift notes from the previous note value (BTW : that's real portamento) ? I guess not.
You can also have polyphonic rising portamento from pitch zero :
Pitch page as in the previous example, but set Coarse to -60ST. FUNs : replace BKeyNum by KeyNum.
Explanation : I used KeyNum here for pitch modulation and again IMHO the Musician's Guide is not correct, I think it generates a value of 0 in response to MIDI key number 0, a value of 0.5 in response to MIDI key number 64 and a value of 1 in response to MIDI key number 127, the key off value is 0.
The effective range is the same as BKeyNum, only it's offset by 0.5. After all the FUN, the offset is canceled again by setting Coarse on the PITCH page to -60ST.
Have FUN !
Wim Verheyen,
Date: Wed, 9 Aug 1995
From: Larry Hall
Subject: solution to program challenge
a recent post (by whom i can't recall) put forth a programming challenge.
the challenge was to generate a program with uses real-time control to change the pitch by semitones of 2 layers of the program. following is a step-by-step description of how to set this up.
- edit program 199
- go to the keymap page and select 152 dull sawtooth (or some other simple waveform - if a multisample is used, shifting the pitch more than a couple semitones doesn't sound good)
- now, do a DupLyr to create layer 2
- go to the Pitch page for layer 2
- set Src1 off
- set Src2 on
- set MinDpt to 550 ct
- set MaxDpt to 1100 ct
- put cursor on DptCtl and press [118][enter] this selects FUN3
- press edit to go to the FUN page
- put cursor on the FUN1 Input a field and press[6][enter] (data slider)
- put cursor on FUN1 Input b field and press[162][enter] (-0.50 constant)
- put cursor on FUN1 Function field and select a + b
- put cursor on FUN2 Input a and press [112][enter] (FUN1)
- put cursor on FUN2 Input b and press [112][enter] (FUN1)
- put curspr on FUN2 Function and select a + b
- put cursor on FUN3 Input a and press [234][enter] (0.74 constant)
- put cursor on FUN3 Input b and press [113][enter] (FUN2)
- put cursor on FUN3 Function and select Quantize B To A
- the FUN page should now look like this:
Input a: Input b: Function:
FUN1: Data -0.50 a + b
FUN2: FUN1 FUN1 a + b
FUN3: 0.74 FUN2 Quantize B To A
FUN4: (all don't cares)
- now, do a DupLyr again to creat layer 3 as an image of layer 2
- go to layer 3's FUN page
- put cursor on the FUN1: Input a: field and press[1][enter] (ModWhl)
- go to layer 3's pitch page
- change MinDpt to -550 ct
- change MaxDpt to -1100 ct
- finally, go to the COMMON page and make sure Globals is set to Off.
- save the program, and try it.
the control slider will change layer 2's pitch in semitone increments from the tonic (layer 1's pitch - the key you press) through a major 7th (1100 cents up). similarly, the mod wheel will change layer 3's pitch downwards in semitone increments form the tonic through a major 7th.
as the challenger stated, this could be handy for ear training. as another twist, if the MinDpt values for both layers 2 & 3 are set to 550 ct and the MaxDpt parameters for both layers are set to 100 ct, the program provides a handy tool for learning about triads. being able to shift layers 2 & 3 up through a major 7th from the tonic allows building any possible triad for ear training. although i must admit, it seems easier just to play the desired interval on the keyboard.
anyway, there's a solution - and one that does a fair job of demonstrating the power and usefulness of FUNs.
i didn't provide a detail explanation of how this works - if anyone wants an explanation, i can post one later...
--larry
Date: Thu, 10 Aug 1995
From: Bill Simpson
Subject: solution to program challenge
This set of functions will change the pitch in discrete steps through a major triad as the Data slider is moved. An additional feature (bug?) is that it gives a minor 3rd and a major 7th for short portions of the slider range.
- On the PITCH page:
- Source1 -- FUN4
- Depth -- 1200ct
- Source2 -- ON
- Depth control -- FUN2
- Min depth -- 0 ct
- Max depth -- 100 ct
- On the FUN page:
Input A Input B Function
FUN1 0.50 DATA warp1(a,b)
FUN2 FUN1 ON a AND b
FUN3 *** not used ***
FUN4 0.40 DATA quantize B to A
How it works
Source1 on the pitch page moves the pitch up by variable amounts up to an octave. FUN4 controls the pitch. Quantize B to A turns the continuous slider into a stepped control, with the number of steps controlled by Input A. For a value of .40 and a unipolar input on Input B, there are 4 steps, including 0 and the ending value. This means the octave is divided into 3 intervals of 4 semitones, i.e. pitch shifted by 0, 4, 8, and 12 semitones. This gives a triad with an augmented 5th.
Now we just have to lower the augmented 5th by 1 semitone. Source 2 is set to a maximum depth of -100 ct (down 1 semitone), and controlled by FUN2, which is controlled by FUN1. The warp1 function with an Input A of 0.50 turns a steadily increasing unipolar source into a value that increases from 0 to 1 in the first half of the range and then decreases back to 0 in the second half of the range. Add the AND function in FUN2, and it becomes a value of 0 in the first quarter of the range (where FUN1 is less than 1/2), then 1 from 1/4 to 3/4, then back to 0 from 3/4 to the top of the range.
With a depth of -100ct, this lowers the augmented 5th down to a regular 5th, giving the major triad.
The bug/feature is that the points where FUN2 changes aren't exactly the same as the points where FUN4 changes, so there is a region where the major 3rd is lowered to a minor 3rd, and the octave is lowered to a major 7th.
Programming challenge:
- Can this be done with less than 3 FUNs?
- Can it be done without the extra minor 3rd and 7th?
Does anyone know the exact points where QUANTIZE changes values, for different numbers of steps?
Bill Simpson
Editing Tips
Date: Fri, 8 Dec 1995
From: Ken Shvetz
Subject: Sample editing tip
I noticed in a thread by Joe Albano that he had problems with a "crash sample that got cut off too easily as it rang out" because of note events being used up. I know nothing about what you guys are talking about
(i don't even know what "gating" means), but would setting the sample's "IgnRel:On" (on the Sample page) fix it?
That reminds me of a technique that i might as well share. All you K2000 users should already know this, but judging from the crappy samples i've retrieved off the Net, most users do not. It concerns using DSPs to edit your samples to make them sound right--specifically, applying a volume ramp curve to the end (or beginnng).
If you who are reading this already know what i'm talking about, just delete this from your mailbox and move on, cuz that's all the rest of this is about.
For almost every sample you record, you should be using DSP functions to edit them (unless you want them to click like the ASR-10 samples i've heard). For example...
If you want to have a sample fade out naturally, use DSP functions to edit it. Again, this is all in the manual, but since i'm sure 90% of those on the list have never opened it, here's what you do (say, to make a cymbal sample fade out):
- Go to the EditRAMSample page (press Master, Sample, select your Sample, then press Edit, then press DSP).
- Enter 11 on the keypad to select (or just scroll to) VolRamp.
- Enter 999999 for Start to get to the end of the sample. Move the start point back a little, about an inch or so on the display.
- Set the second dB value to -96, leaving the first at 0 dB.
- With the Type of volume ramp set to LIN (for linear), press Go.
- Audition the sample. Notice that it nicely fades out without ending abruptly!
This is just one of the techniques anyone who records samples themselves should be using regularly, along with other DSP functions like Normalize, Truncate (i hope we all know that one, although, like i said, from what i've seen on the Net...), Resample, VolAdjust, etc. If any of you non-manual reading users found that to be something you had no idea you could do, let me know if you'd like to see more sample editing tips such as this one. If you're all just mad at me for filling up your box with stuff that's already in the manual, tell me to shut up. I believe i've almost mastered the Sample Edit page and features (i create rap jams, so it's my most-used page), so if any of you have questions, feel free to ask.
Ken Shvetz Back to Dead Letters
| Main Page : | Documentation : | FAQ : | DSP Blocks : | KDFX : | Algorithms |
