X-Treme SFX 15 [Bug Loop Soup]

OK, put away the magnifying glass; this is serious. We're going to have fun with a little bug I discovered in V2 that pops up when automating the Arp. It requires a keyboard controller that transmits true MIDI Note Off messages, and it only shows up with note messages played below G2 (using Project5's note convention). The other requirement is to use a velocity message to trigger the Arpeggiator's Latch function via MIDI Remote Control. The reason for doing so is to add a natural breathing technique to the arpeggiated patterns that can be controlled directly with playing technique.

So how do you find out which flavor of Note Off message your controller features? If you're lucky, you can decipher it from your KB documentation and MIDI Implementation Chart. If not, get some visible feedback from a MIDI monitor app, like Bome's postcardware: SendSX. Open it up, set the MIDI Input to the device under scrutiny, and play a note on the KB. You'll see a couple of sets of three hexidecimal numbers each. Here's an example of what you might expect to see for a medium-velocity note at middle C:



The 90 indicates a MIDI Note On message on MIDI Channel 1 (the zero part). It's been transmitted at the hex equivalent [3C] of middle C [C5, or MIDI Note# 60]. The third number in the set is the velocity value. In this case, it's a velocity of 87 out of 128 possible values. The second string of status bytes shows a Note Off message, indicated by the 8 (Note Off) and the 0 (MIDI Channel = this hex digit +1). It's also at a middle C note value (of course), and has a release velocity of 26. Now you know what that Dimension source is all about in the MIDI Matrix, and not all keyboards feature this. But you'll also see why the velocity messages can't be separated from the note that they ride in on. They're inextricably tied together, although velocity can be copied, and/or used as a source message by MIDI Remote Control (without being "intercepted").

The Note Off message cuts off the companion Note On message; otherwise, a note would drift on forever without any resolution. This has nothing to do with release times on envelope generators, or anything that's synth-specific. It happens the instant that you lift your finger from the KB, and measures the speed in which you do so. You might instead see some messages like the following examples:



In the second and fourth examples, the manufacturer has chosen to use an alternate version of the MIDI spec. That is, instead of using a Note Off message to stop a note, they've implemented a second MIDI Note On message at a velocity value of zero, which effectively shuts off that companion note before it with one of a velocity that's inaudible. It works well in most cases, and many KBs use this technique. All M-Audio products come to mind.

The third example displays a message from a KB sending a static velocity, like you might use for a drum pattern or a "glue loop" that anchors your composition's background. You'll see that the velocity in Ex. 3 is set here to send a maximum value on each note, and the the shutoff message is set to a velocity of 64. Note that this cutoff velocity value is just over the halfway point in the value range, so anything that was MIDI-controlled by velocity could be "turned on", but not off again. Keyboards that are incapable of velocity-sensitivity are still required to send a Note On message with a static velocity value of 64, or you wouldn't hear anything on the receiving end [Example 4].

OK, so now you know what you're transmitting with your KB. If you have the Note On/Note Off variety, you can now load up any synth into a track and play along with our game here. Any synth will do, but keep in mind that many sample-driven engines won't have the raw material at these extreme KB ranges to work with, so I'd suggest a VA or similar synth with wide range oscillators. Normally, the MIDI-controlled Latch function would latch on with stronger velocities, and release the Latch with a softer note. But ...



1). In your loaded synth track, enable the Arp at default settings, and right-click on Latch. Select Velocity as a source, and OK out.
2). Open up the Editor for some visual feedback, and play any note from G2 downward on your KB. It's got to be live input, not a pattern or the preview keyboard.
3). Listen to the stuck notes play back endlessly. You can keep adding to those in the given range for more complexity, and you'll even see them "stick" in the Editor's keyboard as "blue notes". Add other notes above that range to modulate the evolving "pattern". Triggering the same note keys from your KB or any other device will not stop the latched notes, but it may modify the timbre, depending on the synth patch chosen and it's polyphony settings.


Here's the cool part: All of the Arp parameters are still active, so you can change the Pitch Offset, Shape, Octaves, Rate, and everything else and "play" the stuck loop. It'll persist through patch changes, but not through a Replace Synth action. You're not going to destroy it; even an All Notes Off [Panic] over MIDI only dents the loop, muting it temporarily (as another compositional device). The only general way out is by double-clicking on the CPU meter, or by pressing Play then Stop [double Spacebar] in succession [in this case: effectively the same action]. Unfortunately, this can only be recorded with an outside application, but using something like Sound Forge's Remote mode will capture the moment for use later on as a wavfile.

Since this bug loop inserts itself into the Arp stream, you can play other notes over the top and have those slide in-between successive iterations of the bug loop. If you've got five notes bugging out, it'll carve out 5 slots from your fingered input. Now this is also tres cool. Mouse the Transpose slider in your track over it's extreme ranges, and the looping bug travels with it. Incredible sounds result from mashing this loop up against the extreme boundaries of your oscillators. Add some audio effects to the track; you've got a new sound source to play with.

My best guess as to why this happens is that the Note Offs are somehow lost in the Arp or Project5's engine, but why only in this range? I used the Transpose track function to try and work around this thing, but that seems to actually modify the incoming MIDI Note values and shift them to the problematic range. Hey, I'd rather just roll with it. I kept having this anomaly pop up in my compositions, and now I realize it's a direct result of my technique. I treat the octave switches on my two-octave KBs as I would any other note, used as frequently as anything else, and that caused me to dip into the "capture range" in the course of my playing.

Hey, it's not your bread-and-butter technique, but, after all, this is the X-Treme "Special" FX series. It's small point of contention here, but a sound effect label to me implies a Foley effect of some sort. I look at these techniques here as more of a special effect. As a parallel to the graphics CG departments in film, sometimes the best special effects are those not immediately obvious or invisible to the naked eye.

Whoowee....radical SFX there, Tom! Is it a bug or a feature? -from another thread

I was going to coin the phrase b-eature, but I like fug much better. You know, it's a bug, but an enjoyable one. A fun bug, as it were.

I knew that if I played around with this long enough, I'd have to amend something that I'd stated earlier. I said there were only a couple of ways to stop Bug Loop Soup from a rolling boil, but I've found one more. If you're persistent enough, double-clicking on the stuck "blue notes" in the Editor's preview KB will stop it, or at least slow its ass down a notch. Now this isn't reliable, and might only knock one or two notes out for a lot of effort. My theory was that if I throw enough notes at it, P5 will give up, and that seems to be the case here. It just seems to react to the assault with a random surrender, with pockets of resistance popping up in unexpected places.

I didn't want to leave out the M-Audio users from all the fun. I guess that I should've found out how the Pro-88 treats Note Offs before issuing such a blanket statement. But Trigger Finger users can exploit a fug of their own. Let's face it: it's not as cool as Cakewalk's bug, but it'll do in a pinch. You need to drag out that Enigma software, and bring a different MIDI Note On/Off Toggle message to each of the pads, to prepare to make the Bug Finger Soup dish.

Now this Note On/Off Toggle message is supposed to send out a MIDI Note On with the first hit, and shut down that message with an MIDI Note Off of near identical values. The theory is sound: send out one note; shut 'er down with a second press. Guess what? The Note Off message didn't get the memo. It never is sent after the leading Note On, so the notes drift onward mercilessly.

When you run these ghost notes through the Arp, Project5's little fug isn't activated; evidently because there's no Note Off message to lose. But these note clusters can all be affected by the same Arp, track, and FX parameters that are detailed above, so you can obtain similar results. One neat variation is to program two adjacent pads in the TF to identical notes and channels, but have the first contain a Note On/Off Toggle, and the second in the pair with a plain old Note On/Off message.

When the first note becomes stuck, the second pad will allow you at introduce a chunky rhythmic quality to the ghost note. It's almost like palm-muting out the ringing notes of a guitar, and really adds a funky snap to it. You can have the eight notes of a one-octave scale programmed in, and have a "shuffle pad" right next to each note. The close proximity gives you a chance to slide in some polyrhytmic phrasings into the ongoing Chord of the Spirits.

Say, first the invisible aftertouch; the PSYN "anomalies"; the Send level remote control; the Arp blips; and now this. I won't even bring up the Groove Player, and the associated abductions. Is Project5 ... haunted???

[Now if that isn't a setup for your Halloween challenge, Jace, I don't know what is ... Take it and run with it.]

rabeach try the voxengo recorder. free and works great. :-)
yea i forgot i had downloaded it and failed to install it.
why are you talking to yourself.
this is such a great x-treme sfx i thought it needed a bump.
yea, it is right up there with extreme transpose.
the best always gets overlooked.
so true.
ciao.
alright ciao.

We're going to have fun with a little bug I discovered in V2 that pops up when automating the Arp.

Forget I ever mentioned that bug. It's small potatoes. Yesterday's news. OK, I guess that I should give credit where credit is due: This little anomaly was the inspirational seed for refining the "technique", and provided the general roadmap. A stepping stone.

What's even better that using the Bug Loop is to find ways to slice a MIDI note in half, and separate the Note On portion for the Note Off message. That way, you can choose note values from the entire range [C0-G10] of the MIDI spec, and not be limited to the selection available to Bug Loop Soup.

I've been able to pull this off with just about any MIDI controller, but the exact method will vary from KB to KB. Remember, the important thing is to interrupt a MIDI note(s) after the Note On, but before the Note Off release of your fingers. That may help you out in customizing this method to your own controller. Here's one way:


1). Load up any synth and patch, and enable the Arpeggiator at default settings. That includes a MIDI Process channel setting of Channel 1.
2). Play a held chord, transmitted over MIDI Channel 1. This is usually the default transmit channel of most controllers.
3). Make sure that the Note On segment of the chord happens in MIDI Ch. 1, and use your controller switches to change the KB transmit channel to 2 or above, while still holding the chord.
4). Release the chord [Note Off] while in MIDI Channel 2 mode, and the chord will arp forever.
5). Repeat, and repeat again. New chords and notes can be added from anywhere on the KB to the "phrase looper", as if it were in sound-on-sound mode.

The difference being that the new notes will continue to build the looping pattern and modify its length, and all of the same controls remain active (as detailed above in Bug Loop Soup). You can construct huge cascading flurries of Note Ons without a way home. If things get too crowded, you can "edit" the orphaned notes using that double-click technique on the Editor's preview KB. As you recall, the stuck notes appear blue in the preview KB, and you can filter any note out of the "loop" with a double-click.

This can get a little tricky, because the "blue" doesn't go away, even after removng the actual sounding note from the stack. The only way to truly reset all is to double-click the CPU meter in P5. But you can eliminate all of the ghost notes one-by-one in this manner, and it's more reliable [100%] than it was in the original Soup recipe.

M-Audio-ites: I didn't forget you this time around. I know that you're (mostly) all stuck with that convoluted series of keystrokes that change MIDI channels. Too complicated. Just play your chord in MIDI Channel 1, and instead of a channel change, hit the MIDI Select (or similar) button to take you out of performance mode. The Note Ons @ Zero Velocity will be severed from their companion Note Ons by entering the programming mode, and hitting the MIDI Select again will bring you back to performance mode. Then you can play along with the stuck notes, and see exactly where you want your next surgical stroike to be.

Nothing to be really concerned about here, but glance over at the CPU meter once in a while. These stuck notes can add up the cycles, and if you go crazy with this (like some people we know), you might start to tip into the danger zone. Just something to be aware of.

Trigger Finger users: you have your own bug to exploit. I having a hard time "breaking" the regular Note Ons, but Note On/Off Togle works well. I'll let you know when I crack this. Novationers: as you know, you're all set. You can even set up the programmable buttons to Notes for editing or sliding in a single broken Note Arp while you're playing. If you really want to get 12-tone with this, MIDI Remote Control the Pitch Offset with something like velocity. And don't neglect the possibilities of programming a KB slider to change Shapes.

Have fun ...

think you need for these is a summary section for each one.

I know that these things get complicated, Tristan. I just can't leave out any information; thinking that someone might use a particular tactic to take this in their own direction.

But I have started to use the bulleted recipe formula. I was hoping that those who wanted to skip the lengthy explanation could go straight there, complete four or five easy steps, and get to the heart of the matter right away. Then it's easy to get up & running, and the rest of the background information is still self-contained for later use.

I guess that tactic isn't working. I'm constantly trying to modify the presentational style to make it readable & easier to implement, but not at the expense of getting everything across.

Back up and punt; I'll try something different.

a little bug I discovered in V2 that pops up when automating the Arp. It requires a keyboard controller that transmits true MIDI Note Off messages, and it only shows up with note messages played below G2

Uh oh. It's not just under those conditions anymore. I was programming some PSYN patches, and was able to "stick" the entire note range of C4 to D5 inclusive. That was a little more bothersome, because now we're getting into commonly used territory.

No arpeggiator involved, and not much MIDI Remote Control goiing on. Perhaps coincidentally, I was working on using multiple LFOs for oscillator modulation that simulated arp behavior. While there wasn't much MIDI going on, there was one parameter controlled by note velocity.

I wonder if this is where the problem starts. Maybe the velocity parameter in MRC is set up to handle the zero velocity Note On flavor of MIDI, and it freaks when there's too many Note Offs involved.

Back to the drawing board; I should've saved the offending project. Now I have to replicate the scenario again. Multiple suspects; mysterious culprit; little physical clues; lots of circumstantial evidence. Reminds me of the Groove Player.

There just might be enough material here to write a Soup recipe book. I'm never going to get this PSYN-tology thing going again. It's jinxed.

I wonder if this is where the problem starts. Maybe the velocity parameter in MRC is set up to handle the zero velocity Note On flavor of MIDI, and it freaks when there's too many Note Offs involved.

that would be disappointing and certainly a major bug.

I'm going to take this from Postulate to Theorum; stopping just short of Law. It turns out that it does matter where you put the velocity and Mod Wheel messages. That is: they must be within the same general feature or section in Project5, like within the track controls, or the Arpeggiator, or within the exposed 'widgets'. As it turns out, the Arp just may have been an accomplice, not the trigger man. I haven't tried out all of the possible scenarios ... just yet.



1). Load up any ol' synth, and assign one track parameter (like Level) in the MIDI Remote Control dialog to velocity, and another (ex. Pan) to CC#1 (the Mod Wheel).

2). Notes from E1 - G2 (16 note values) and from C4 - Eb5 (also 16 notes) will become stuck in Project5. This can be visually verified on the Editor's preview KB, audibly with your audio driver, and reflected in the higher CPU meter reading. The 16 notes between these ranges [G#2 - B4] will remain unaffected. Double-click on the preview KB to audibly "unstick" the notes one at a time; double-click on the CPU meter [or Play then Stop] to reset the audio engine and all of the stuck notes. Assignments made under extended testing conditions cannot be "undone" by Deleting MRC assignments, requiring a new project or template load. That is: the behavior will persist after reverting parameters.

3). Controllers with true Note Off capability always cause the problem. Those using a Note On @ zero velocity seem unaffected. [Further testing required.] Audio tracks are unaffected (as this is a MIDI implementation problem).



Substitute any of the following ingredients for line 1). above:

4). Load up a synth, assign only the velocity parameter in MRC to anything. The lower range [E1 - G2] will stick, regardless of any other parameter assignments, or none at all.
5). Load up a synth, and assign any two Arpeggiator parameters to velocity and CC#1.
6). Load up a synth, and assign any DXi 'widgets' to velocity and CC#1.
7). Load up synths in two or more tracks, and assign the MIDI messages in question to any two parameters across tracks.



Sorry, Cake. I just call 'em as I see 'em. I'm not ****ing about it, but I can foresee that someone's going to get the plain doughnut at tomorrow's staff meeting. I really am just trying to help out.

Where's Nick when you need him ...?

good work b rock i have not had the time to replicate your latest findings yet. will do so shortly. i at first glance thought wow this should be an easy fix groups of 16 certainly a hex reference. but then i noticed they were not symmetrical groupings. but then the requirement for the two mrc assignments answers that question but probably leads to a more difficult problem for cake to solve. which brings my mind back to this little ditty, "you want to minimize the coupling between classes." hehe i love that :-) you should be working for someone you have a great talent for this sort of thing. don't accept less than $62,400 to start. :-)

then i noticed they were not symmetrical groupings. but then the requirement for the two mrc assignments answers that question but probably leads to a more difficult problem for cake to solve.

OK, maybe this is something that the code jockeys can chew on.

Velocity as a source will trigger the lower range of stuck notes [E1 - G2]; every time. Adding the Mod Wheel will trigger the higher range [C4 - Eb5] as well. But so will a Breath Controller message [CC#2]. Jump to a General Purpose continuous controller [CC#17] with a velocity source, and the stuck notes now shift to C2 - B4 inclusive. A straight run of 32 values, including the previously unaffected "divider", and excluding the previous high range.

If you give me enough time in that dungeon, I'll get this figured out. Only about 100 CC messages to go, and an exponential amount of possible combinations. Trust me: I'm not really seeking this out, but it's hitting me over the head at every turn.

Case in point: I was about to answer Sean's reverb question, and point him to using the Echo Delay as a BPM calculator. Works great in 1.5; no so good in V2. It seems that at least this one tempo-based MFX does not get a timely tempo update like 1.x did. You have to punch Play/Stop to get the correct calculation; in effect: jolting it into submission.

Time to walk away for a little while. My eyes are bugging out.

okay i tried it with the yamaha wx-5 woodwind controller and no note stuck. checked and sure enough the midi note off is the velocity equal zero method. will try with the yamaha p-80 electric piano which uses true midi note on and midi note off next. may take a day or two or more before i'm able to and post my results.

edit.......

i'm having a hard time reproducing this b rock. i will download a midi monitor today to make sure the yamaha p-80 is sending a true midi note off. the spec says it is but i'm going to double check. what keyboard are you using.

okay b rock i used the midi monitor and it turns out that the true midi note on and true midi note off in the spec of the p-80 was apparently in reference to incoming control midi to the p-80 as it is an electric piano and not out going midi. out going note off is just like the wx-5 and is the velocity = zero type. anybody else out there got a true midi note off keyboard they can use to reproduce this bug. it is easy to test just download the midi monitor and read b rocks first post on this thread.

Now to solve the problem with a workaround (if possible). I already tried an All Notes Off message, but let's see if some other combination works. All Sound Off; maybe an unusual application of the MIDI switches [CC#s 64 - 69] and a Reset All controllers; perhaps even a variation on the Free Range Split Note Soup recipe.

No love as yet. I've tried all of these "solutions", and nothing can stop the onslaught over MIDI. I'm wondering now if I'm barking up the wrong tree, and it's the audio engine that gets sticky; needing a reset. The puzzling part is in the double-click note solution that works (eventually) with the Editor's preview keyboard. I'm speculating that this action also resets the audio engine selectively, but I think that I'm grasping at straws with that statement.

It doesn't make sense. Only one note is removed at a time. Back to the lab ...

,,,,have you,, tested your " fun bug' ,,in rewire ,,,,,,,,,,,

Genius. Absolute genius. Marcella's lost in the ether for a month or more, and then pops in with the workaround that's been driving me crazy. Then disappears again. She's from the spirit world, I tell ya'. I still think that she's hiding out in the Swamp.

It's a *little* more involved than I'd like, but preliminary testing indicates that the bug is not transferred over Rewire. Load up Project5 1.5 as a client, do all your MIDI Remote Control assignments there in Syn:Ops, and this isolates V2 from the bug behavior.

I can't wait to test it out with other Rewire applications, although I'm speculating that using V2 as a client in another Rewire host might stir up the bug again. There's a lot going on between those commas. Marcella, where've you been?

,,,,,,have you tried ,, increasing your "intake of cafe cubano",, perhaps,,,,,, well,, ya know,,, you might ,,,,,"see" things differently,,,,,

I'm afraid to. I'm already seeing double. As it is, I go to the grocery store and they're yelling "No more cubano for you!!!"

Thanks, girl.