DSP Robotics Support

hum finally i guess i was not really understanding..

I was thinking that the idea was to know if the number of all voice is odd or even then all voice get the same mod depending on this. But i guess that it's more that each voice will get a different mod if they are odd or even in their own number.
Yet the poly readout is only for debugging.

Tepeix wrote:hum finally i guess i was not really understanding..

I don't blame anyone for not understanding my descriptions and questions since they are often quite bad/uninformed and unclear

Tepeix wrote:I was thinking that the idea was to know if the number of all voice is odd or even then all voice get the same mod depending on this. But i guess that it's more that each voice will get a different mod if they are odd or even in their own number.

Even so, your solution for blue works like a bitflipper so it might be used for connecting random numbers to my free LFO (in blue)... albeit with some kind of slope/dezipper...

Tepeix wrote:Yet the poly readout is only for debugging.

It's all a bit to advanced for me not having insight into deeper level workings of FS, ASM, audio basics, engineering nor math.

In regards to id's I never considered using id's until I started experimenting with MVs white noise + Adams random asm. Realized there is an opportunity to add a fun feature to my synth without much effort. But I might have opened "a big can of worms" in regards to making it all "fit in" and work a coherent way

Ok I think got all the worms back in the jar and managed to closed the lid...
Here's what I ended upp with. Nothing impressive for alot of you guys, but i'm damn happy with this simple feature since I never designed my synth for it.
Poor cpu...

Multi Voice 3

Ok.. and final sound demo using multi voice with qwerty keys... from me again that can't play regular piano/keys still my laptop annoys me 2023 and can't even hold down certain keys at same time...

First clean --> then add detune --> then add odd/even detune (multi voice) --> then add spread (using random values as modulation to my Pitch Pan section) --> etc --> finally at the end a little bit of use case of Martins recently posted "Cascade Oscillator" that now sits in my Aux Osc...

Multi Voice 4

Wondering...
Does full speed rendering/bounce usually work ok with FS plugins? Or is it an issue?
Maybe mostly if Ruby is between knobs/modulation-sources and their respective destinations?

Wondering if I can open the floodgates (to hell? ) and allow for 128 or 256 voices (for rendering purpose).
Does FS have a "logical" limit?

R&R wrote:Wondering...
Does full speed rendering/bounce usually work ok with FS plugins? Or is it an issue?
Maybe mostly if Ruby is between knobs/modulation-sources and their respective destinations?

Wondering if I can open the floodgates (to hell? ) and allow for 128 or 256 voices (for rendering purpose).
Does FS have a "logical" limit?

A FlowStone plugin is usually capable of fast rendering.

One thing to keep in mind is to always consider the sample rate in critical situations. So, if you’re inclined to use the float value 44100 as a sample rate, don’t do it! Use the sample rate prim because this will provide the correct value/ratio.

Another thing to think about is delays. If you want a 1.5 second delay, the sample rate must be taken into account. So a 1.5 second delay time would be 1.5 x sample rate.

That leads to a point about setting the maximum delay in DSP or ASM delays. You should take into account the maximum likely sample rate. Let’s say the maximum delay time needed is 2 seconds, so at 44100 you would enter 88200. But at 192kHz it would be 192000/44100 x 88200 which is a max delay of 384000 samples. For modulated delays don’t forget about positive excursions which could go beyond the max samples you set.

Something happened to me a while ago, with my SIM-BP. It would only render correctly online (x1). I put this down to having a lot of Ruby-based midi processing before the MIDI to VOICES prims. However, when I tried to reproduce this later, for Discord, it worked fine! I suspect that Reaper’s updates may have fixed it but I don’t know. To be on the safe side, I now try to do voice processing in the Pitch output of the MIDI prims, so the MIDI side is as clean as possible.

The obvious thing to say is to test your plugin and if you can’t get it to fast render then just say in the release notes. I’ve heard of some pro plugins that only work at 44100 (maybe old ones, not sure).

Regarding limitations, I’m not aware of any, but the obvious one is CPU usage.

Spogg wrote:Another thing to think about is delays. If you want a 1.5 second delay, the sample rate must be taken into account. So a 1.5 second delay time would be 1.5 x sample rate.

That leads to a point about setting the maximum delay in DSP or ASM delays. You should take into account the maximum likely sample rate. Let’s say the maximum delay time needed is 2 seconds, so at 44100 you would enter 88200. But at 192kHz it would be 192000/44100 x 88200 which is a max delay of 384000 samples. For modulated delays don’t forget about positive excursions which could go beyond the max samples you set.

Good thing to keep in mind, thanks!

Using alot of MVs interpolated delay and his transition oscillators. But he was very meticulous in his code comments/descriptions with that so, think I took into account for rates above 192000... all the way up to 256000 or 320000 maybe. Only using a very small delays though due to memory usage and cpu cost...


On another note Spogg... while I have you in the thread

I'm in a very weird limbo place with my plugin... it hasn't actually "played" in FS3.08 for a long long time, calmly silent, but without crashing FS when trying to play it, it seems. It's probably on the edge of what FS3.08 can handle in some aspects even schematically.
Having to streamline the schematic all the time Going to reuse more of my knob bitmaps I think, once more, to get rid of some.

So... work on the schematic and individual components separately is done using F3.08 (all put together won't work in FS30.8. They don't seem to handle it).

But is then loaded into a FS4 zoom-template and other missing FS4 specific prim's are "synchro-copied in"... after that exported to final VST plugin.

I'm not an official FS alpha tester so I don't even know what release it's at... Got my current FS4 from Malc initially
Notice now that FS4 files seem "version specific".
Would be nice to perpetually test in later versions of FS4 so I can keep it working as FS4 progress

I have your invite, still i'm not much of an alpha tester since I only use very small amount of prims from FS4 and only it's engine for exports... not that technical for that matter either

Enlist in the Alpha tester army anyway?

Would probably work with more FS4 components in a "Saguaro.one Two". But that's not planned right now

For a very long time I’ve used FS 3.06 following being strongly urged to do so, especially from tulamide. He can explain the tech details better than me. Ruby threads or something.
What I found is that 3.06 is far more stable to work with. Yes, I get a crash sometimes, especially as a schematic grows in size. However, the crash frequency was far greater with 3.08.1.
You can open your 3.08 schematics in 3.06 successfully. The only thing to be aware of is that Martin sometimes uses ASM functions that aren’t available in 3.06. Having said that, all you need NOT to do is edit any of them in 3.06 because the code is somehow pre-compiled from the ASM text when you open it in 3.06 (I believe). If you change anything in the ASM text it’s possible it may not compile correctly in 3.06.

I gather Maik wants anyone to do anything with the alphas, simply because the more people using it, the more likelihood someone will find an issue. So you could try it and see, but of course if you save in an alpha you won’t be able to open it in an earlier version. I keep any saves (Save as…) I make in an alpha in a separate folder.

Another thing to consider is that when a schematic is saved, the save takes up more memory. Most programs seem to wipe the current version when a document is saved, so you can’t Undo. Also, when you change anything, I believe it keeps the previous version in memory (maybe as a delta, I don’t know). That means a huge schematic will take progressively more memory and I’ve had “Out of memory” messages at times. My approach with huge schematics is to Save As frequently and close/open FS often, once I get into the “crash zone”.

Yep, I remember you warned me about 3.08 from the start, or rather recommended 3.06...

My schematic was already spaghetti by that stage so I didn't wan't to add another source of failure "if" there could be one.
Did work with 3.06 and also had separate saved files (to be absolutely sure). But the actual issue then was due to CPU and crappy(ier) schematic at that time. Had not moved to using that much of MVs asm's either

As long as my synth flies in any of the FS4 i'm happy. Much more concerning is 3.08 (haven't checked with 3.06 recently) starting to buckle under pressure.
Telltale signs seems to be FS's pauses/freezes become longer and longer when copying modules (even simpler ones). It's "warning" me... as you said, entering the crash zone... I rarely get crashes though. Maik has done a very nice job with FS, it is quite stable compared to alot of other softwares...
A strange encounter as I wrote some time ago was when I started loosing prim/module-connections when copy pasting stuff. My fsm files size was only around 4mb I think... Perhaps it's just "that" condensed, the way I have so many intances of everything? No clue...

Spogg wrote:I gather Maik wants anyone to do anything with the alphas, simply because the more people using it, the more likelihood someone will find an issue. So you could try it and see, but of course if you save in an alpha you won’t be able to open it in an earlier version. I keep any saves (Save as…) I make in an alpha in a separate folder.

I'm not brave enough to move my project to FS4 permanently yet, until Maik maybe has "locked in" to some kind of stable version... in the sense of allowing for opening/exchanging files between those stable versions just like 3.06 vs 3.08.
Have to consolidate my schematic a bit so I can paste in separate parts of the schematic into my FS4 zoom-template.

I've reached the upper limit of what my CPU can handle so can't add much more polystream complexity.
But the schematic itself will grow quite a bit when/if I start making/adding more Osc modules, more settings for those and also adding LFO's (iow things that are "switched between" if selected).

Haven't "stolen" any solution from anyone yet since my Osc modules are quite simple. But I am ogling a bit at your wavefolder solution, that would make for a a nice Osc Module
Also creating a "broken saw" Osc Module would be cool if I can figure out how...

Of top of everything i'm kind of making room (schematically) for creating another version of Saguaro.one, but with a more regular osc section (but simple 3xOsc layout).

Btw
Sorry for all of my eeeeendless babbling questions/posts/answers. Thinking out loud alot

Quite some time ago since using Discord or even Slack. I'll send a request to Maik when eventually on Discord then...
Hoping my schematic works in newer FS4 alphas

Just me, my qwerty and audacity again...

Clicks, clicks.. clicks Seems hard to reduce when modulating some destinations... LP... or HP filter to subtract mod wave or something... Better than slew limiter...?
Any tips?

Multi Voice Pancakes n waffles
No delays, just multi voice, panning and modulation with saw and a slightly custom adaptation of MVs Random S&H.

Reminded me a bit of the latest Tron movie. Man, I wish they made another sequel... with more Daft Punk. Loved that soundtrack... completely made the movie...