Mystery Blip

[dmbaer]

I have attached a simple MXXX preset that illustrates a problem I’m having. Maybe I’m just asking MXXX to perform an unnatural act, maybe it’s my error or maybe it’s a bug.

We have Mod 1 which has a basic upward ramp as waveform. It modulates MP1 which has a basic upward ramp transform map. MP1 modulates MP2, which has a cyclic transform map (there should be no discontinuities in the output when Mod 1 is running). MP2 modulates the volume of a utility module.

It will appear in this context that MP1 is doing nothing useful, and for purposes of this problem demo preset, that is so. In my real-life preset pursuit, it needs to be there.

Here’s the problem: run this (no input audio needed, there’s an oscillator module creating sound). At the Mod cycle restart, there’s an audible blip – sometimes louder, sometimes hardly discernable. But it shouldn’t be there at all.

My appreciation for anyone who would take a look to see if you spot the cause of the glitch.

Code: Select all

$eNrtWltv2zoSfu+vEPSwL2cTk9Qdm+TAceI2QNzkxG6TfWQlJtZWFlVJzqW-fmdISZZkO+vW6QIHaIra4kdyhhwOP85QPvrzeZEYjyIvYpkem-SQmIZIQxnF6cOxuSzvD3zzz5N3R5O7u7tClCXAhTGSScKzQkQ38cO8vOapSI5N6DjKZVFIEDaRkTg2mR2dDmRYtipmLxlUXMap4PlBNueFgMplnou0vM4FKLg4OzZnoiiN0yTOTOM9j1OiZEOfr-rhZpolcVmKfBp-F-TYDAK7j7Jj07Jc05jw58vx1Wkiw68IA8oAXCZlfM1zvhDQvGgmo6TDZJS+k6NmzMOUJy-fRW7UD6M5T2HKepKT+Nk0zlP+JUEJtH6+nYv0QxxFIlVipwspy3kqigKKh8QBRC7zEEdkGrdxGsknbZoPIBqUgzHu4wSGB-Y2znL+BN9aHw7tWsZpaYwT-lCcgk4btD7j2lmEEhYEgeO65mBbM7vfbNBWhqplCOOUOelMaxwniXqYzuUT2Ezmn3myFIXCZvM4VKszaPWnhmoGfkACl5Dx2NQAayONCjDSFRg7jyNRddsml9VyA9chXbkr5CfkWr9Irl3LJcSzArctd4XsLncqU-kAztvuArBqXGjJ6MHVX6WL1ohP2tpxTs14LETG49bM7WMTywpBR1nbEkAMX3gaEWMqEhGWIrpOlg8xuvw-Kfjp1bLMlmXHjwY1SjeibCNqbUTtjajTQ-WIivphw1adXJxdXKTQVRl-vEySj0ANx+ZVEYLP81LmBrRFYlohUMYeF2kknsHaB7QCpqXgiSasCS++XqRKj35UdU3VFeqDur4uWH41j1o0NL9Tn--G5vU0pjUVZzWPjYBCC9Wwgc7yly4wzsW3JbD7S58sVk2uwpI-ip6g67gM511oKhZxKdN+y1uBVjx0XGbZ1GIO9Zx2bRyVc01h0znPhKEp7aPMFzwxjUcklGPTgerz9FEkElr8T+6zNjEdUUxHDhlof6kdYUsbT7c52NjI0syqvL8eEwy+FJlxUYoFUSdmtdEQoCiog7B+E2utid1v4qw1cftNvE4THNCZyLRxVQnbjOQyheXw1ebIRM5LOONhJ8xynhb3YPRdTpat5whdmXbQSDw5Gi7wHF5G4oe0oMAD+spCrOtcVwRge5oD5WMnR7i-bzC2UGddZ-s0wE-Qw6cyBv0vNTdUxbcmhkbsdlawXmGFYRR1N+iwLLk6Sg4JbcGnEImdp8sFHM0tdAQrlnf7K6gVMtF+Ha6ylnQp7ss-VIRoHPzjofwX7PZU-gGTzYVss9RIxWd9RQ0MIxbpUi2qlnvgRqfGwJCKqTb12DK8pv6vJUebKqJq14uEv7TP4Y0124RjkwlweFyIUKZRsUHCjchkXl7yUlPwWv2Ug0v3GbVmnY0D61VuGdv5c9br9pxBsK0j-ueN+GZt-eot+jBo7-a8gHHCzMEhNuHKR7oVl-KhC0x4Od84qFbFluFkPE34E-Cw2wUlPPKusBpN1EgP6IaqXA+2XaVyCdo7MvGBdbEbCJJgn6HftbvfTKYbZ7bCt0wMGnyQyXqvS5E+qKOgu8encQKOJ67B074AB1zd3-eO9F59bw2nTzzDFdywWp9lslyoefm258DR7xNGvLUGev3pGn5Tm3TQ4uNBE7gNVIgJoSakclCEz15msopqwLCuEzi+GxDPhqywG1+ogyKGOEnD5ynkcaGIkD4bv4uLKhEdtFDF+jc8fYBeNuaIFa4pWhXxGPis4xdSPZ2CEZeZZpk64HmrA7c62gbNULrHXBM+jcFz5FOdkMOuAv8t54s4hFnk8j8Qs6vxVp45ixeiymo5UG4C45uoHBmsLkBiBIUQYtQJqFBTxIdOzazOdcZj167zBtV0ZgwTSPiVOdqZCTbFP6KyGHVWPgGv43ne6sL6XVbSB-UQh9GjWk+je8bpkp7WmBclIMtSVjOeZkJtu8PmuI+mmHnNFYUBeC3412aTdTdwVZpUZFcXW4q2mdVY26SN0vNvbS5o4Pf1eBpkwp-HmEjr+K+F-lWdayssTlXLFgR+IlM0lgqAcavO5rkQ13ghM66vAtQIEC-m2gCHFu6hpdD2oLoAU9GJyWhZlHKhrb2zs2+Nztddfk1+rbJeyl+ms60ANvJqqU6O3uc8m58cZUpSe+W+VZalaNxS33rJp+lcJPcqdKOHHgtcQm0G-1QYh4P9wPOFTOOwQOVK6I6y0U2VXHZIHZ96zHeYT99ULrAqcCpItfwd5L4mywKXdyzqeASyzzcco3Vo25bvezBCh+0v9wMcS6sFsw59j9mO77gw7FeFDyqnGObgNXgFGS-i7+JGPhUNe7ddaFDzVifFRDZvWHwYYvSrDuxpKTOjAA9PIPSt8BRhnpcr-Gx6MxOLTE5f0vDYhI5mmxRZXVIeraZ-JkL+UpORKrTrEq7D1YYILxYLEcUQyK4YscOOrCmtxOgr0JpXprCzak7rDRUTVLEAov99R-D7juBX3xG0XQ1jida2g3Euo1g2brZLwNQ+L1Xvv1cINejM+OQIYt5ILqotqDdwJ3DERcC7QWS5Jgy-EaF6rdHEVrh7gKC+xEn8XRm+JppOcKFWbBo-pDwx6hFU-krNX8wCVBwQSl5xrZoIguZvO2n85oO-852hckBYLjiOjDpGZ6ozJp76q4BkdCTTMofsCvbNAIofZSl6UO8940kPIMb7TxfnUYxk0HFugKfQBAIGfSfYqdQpJ+ymB2FQ4yKN4lyoKGC131aJ6KHrup7FKPUcy65gLe5a5CG+MDB-MMtl+P6Pt26j2mnv-yHJHfRs2Dcq-Smj-lrLnQKJwlodGM2NMTyf8iIO4VvfgLxiVYi77W2Wbd4UfkrjsoAg1etB+M4Ik8aUZzP5HuC76iXSuu2b8Ey-h0YVarUVfEe3B0g7dNTcSWzqu9Wfp5YZTwHbCmzfdiyX+Hsr8Jlt1X+1Atv2aQAwDcCOeyqgVawIT4QyAqlbQJ2V0FDxq2fZHoP-TuDjqQLJ0g8r61Zp1a7btx31PBiQQ1ziWfvOjBHf7VmOQbpDLNe2PObsLd4htBJr+Yw41GfgC2xvsWD+nlEcYqNBmO-Ze48abMF6RgEjuQElgQXMsLd4r-Ymz8Vs2A9sz9pbqk3XHMX3CXNdn1CXBnuLd-y+TQLXDXwPNhpS5Z7iX8-FdhezsoK3GqUTBDQgzN97tzi+1ecZ3wscx3Ydx997BV3glcYxYOE8Rlzb25sd3XUGcS1ksLfxZo-4pGcT2IvMdwgS5d4m9xzCGgbBc+Rt3NkLqN0zCvOoV+3Gvd0Z-MTvGYV6rue5AbOIt7fNFcmtDqTKwfe2dUBdq39OwzHqvNUmh53i+HDgOC6xV3GAbbHAD-DNzZbocEf5u8aOdA3BsB6v1B6FfgVkYCBZXW7zUjzI-AVTpIv0XuI7hDmm0aZxPuFxgg+I4-ft7S18DVadIFovYSzXyzyThVApelHmEPal+K4GZ4ON5zCMEEYBz2ra6v65K6H6pWSVBJi-QMOgUoG5UssSaJg8g2Cbl3n8fOL5Z0NwEQhuiDO0fOKNRsT2Iascec54DJk0oWcWFBkFE+suR-qHRT0BQ+hiYSgF60+HI-VbMzI8rbp2uhx9hkQ6hPC5K8Ie2WQMfOCTEfQ6D6quvcZHOpPbqWunKf6K74u6udhR8Xp7sCSY7l1zDUTxxofhh4UfNn44+OFO1MJNBvodw0T-pA2-1O+mJvjTP-zNqNH+vWZ1CzOom+ifiTxhada+HdK-6vscFzFkGFUndVO0ft20qdWG+yRny31SrztOJcS77pYiAgfaKy13UbZZhJ79rHsttq3ZLmo298c3DnSGC1OJUAVEWRtlNWq1UatG7TZq16jTRp0adduoKtR3du-gqfXb7JN3-wXXsQNy

[werzel]

I looked at this from an unskilled perspective (I wasn't sure what you were trying to accomplish.....might still be the case) and have one question which may or may not be helpful.

Is the blip caused by the retriggering of the oscillator? I noticed a version of it using all possible combos of the MPs and MOD. What it reminds me of is a wave triggering at different phase points (non-zero crossing). Would that be your culprit?

OK, not retriggering but being at a non-zero phase at that point in time.
I recorded a segment and looked at the wave. Each 'blip' occurred at a non-zero crossing point.
I couldn't find a way to force the Oscillator to begin which made for tuning it to an even number of cycles per bar a useless endeavor.

[dmbaer]

I looked at this from an unskilled perspective (I wasn't sure what you were trying to accomplish.....might still be the case) and have one question which may or may not be helpful.

Is the blip caused by the retriggering of the oscillator? I noticed a version of it using all possible combos of the MPs and MOD. What it reminds me of is a wave triggering at different phase points (non-zero crossing). Would that be your culprit?

OK, not retriggering but being at a non-zero phase at that point in time.
I recorded a segment and looked at the wave. Each 'blip' occurred at a non-zero crossing point.
I couldn't find a way to force the Oscillator to begin which made for tuning it to an even number of cycles per bar a useless endeavor.

The blip happens (always) when the Modulator transitions from its end-of-cycle to a new cycle. Apart from that, I'm clueless. To tell the truth, I did not record the output to examine later. The blip is audible, and that's enough to denote a problem.

A discontinuity at some point in the chain probably is the culprit, but I can't figure out how to track it down.

I do not believe that the Mod has any control over the oscillator behavior and I'm pretty certain discontinuities in the audio wave input is not a factor.

By the way, since you asked, this preset effort is in pursuit of doing something inspired by Waves Brauer Motion: an orbiting sound source. The smooth transition map in MP2 is intended to control gain and optionally bandpass filtering as the sound source traverses its orbit. I've got another map for pan angle, but that wasn't relevant to demonstrating the problem.

[werzel]

I looked at it again.
If we agree that the Mod is the culprit, then I would suggest changing the shape to a triangle wave and then adjusting your LFO rate if necessary.
Even adding smoothing did not help to mask the abrupt change from full position to zero. The blip did change, but it only changed in tone.

Circular auto pan.....perhaps even a sine wave then to make things even smoother?

I'd like to hear, see, try out your final result!

[MeldaProduction]

I believe it is the saw - it is interpolated whenever needed, that means that the discontinuity is interpolated as well, so in a way there's a quite jump from -1 to 1 through values in between. Sometimes it doesn't matter, sometimes it does. Try using a triangle instead as werzel suggested.

[dmbaer]

Try using a triangle instead as werzel suggested.

Remember that "barbershop pole" preset you demonstrated in one of your tutorial videos? What if somebody suggested "try using a triangle instead" instead of the ramp waveform there? Would that have gotten the job done in that case?

[MeldaProduction]

I'm afraid I have no idea what you are talking about . I didn't do most videos really and there are sooooo many!

[dmbaer]

I'm afraid I have no idea what you are talking about . I didn't do most videos really and there are sooooo many!

Then you should watch it sometime. It's really quite an excellent tutorial.

https://www.youtube.com/watch?v=fjZBjbsxCEs

[MeldaProduction]

Hehe I watched all of them, I just don't remember . What should I focus on here? (time)

[dmbaer]

Somewhere around 8 minutes in we see a modulator with a ramp wave introduced. My whole point was that there was a suggestion earlier in this thread to use a triangle wave instead to solve my discontinuity problem, but a triangle would not work any more than it would work as a substitute for the ramp wave in this tutorial.

Now I understand your explanation as to why I'm having this problem. But I'm not arguing that things should work differently or that something needs to be fixed. I'm just arguing that when you need a ramp, a triangle hardly is an adequate substitute.

[jmg8]

I took a look at this today and found out what is happening.

So basically it is the transformation shape that you have in MP2.

In a normal situation, lets say you are modulating volume with a saw shape. When it gets to the end of its cycle it has to VERY quickly jump the value of the parameter to the start value.
But by adding the transfer curve you did. You are asking it to very quickly go up in volume and then down again (between the cycle end and start)
So actually it is not a click or pop type glitch, but rather the exact volume increase you asked it to do with the transfer curve.

[dmbaer]

I took a look at this today and found out what is happening.

So basically it is the transformation shape that you have in MP2.

In a normal situation, lets say you are modulating volume with a saw shape. When it gets to the end of its cycle it has to VERY quickly jump the value of the parameter to the start value.
But by adding the transfer curve you did. You are asking it to very quickly go up in volume and then down again (between the cycle end and start)
So actually it is not a click or pop type glitch, but rather the exact volume increase you asked it to do with the transfer curve.

Actually, that's not the issue. The volume curve (actually it's a distance curve) has continuity between cycle-end and cycle-start. We basically just want to smoothly traverse this transform shape, and the ramp does that nicely except at the point of cycle-restart.

The calculation of the current ramp value is done with an interpolation. If that involves an interpolation between max and min values when there is a transition going on between cycle-end and cycle-start, we have an unpredictable result. This is going to cause the retrieval of a value from some unanticipated point on the distance (volume) curve away from the intended cycle-restart point. Since the distance is at a max (and volume at a min) at this point, any other value is going to be louder and thus is noticeable as a blip. The fact that the blips vary in how audible they are illustrates the random nature of the interpolation.

A solution is to take the final shape and use that directly in the modulator, skipping any intermediate ramp shapes. Several features must be dropped when doing this but the blip definitely goes away. Once Vojtech mentioned the interpolation, the cause of the problem became perfectly understandable. I cannot think of any workaround that retains the ramp as part of the design.

[jmg8]

I disagree.
Going from max to min has to be interpolated in order to not create click sound.
Therefore is is very quickly going from max to mix. But it has to follow the transfer that you set as part of that. So halfway through the max to min transition, you have asked it to go up in volume.
Try disabling the transfer curve and you will see that there is no click.
BTW for me there is no change to the volume of the click, it is the same every time. Have you enabled the sample based thingy in the settings?

[jmg8]

Try putting the modulators smoothness to 100% and it will slow down the click. That should make it obvious what is happening.

[dmbaer]

Try putting the modulators smoothness to 100% and it will slow down the click. That should make it obvious what is happening.

Will try it, but I don't expect this will help things ... and might even make it worse. Will report back when I have time to try this, however.