Prototype - Free running poly oscillators

[adamszabo]

Thats good exo, although it still behaves strangely.

I have made a video to show you guys the problems. To me the standard/perfect free running phase is like in many synths like Massive, Rekator 5 etc... they are all doing the same thing so to me thats the correct way.

I demonstrated this using MonoPoly which uses the same method. I have two sawtooth waveforms slightly detuned. Keep an eye on the FFT spectrum, and one can see the phasing effect. There are two things to notice: the phase starts and resumes when the key is let go and pressed again (so we now know it only counts once the osc are active again, its not counting in the background), and the second thing is that, when the osc are detuned, and when i set them the same frequency, they will be drifted apart since the phase difference, however the phase difference stays the same. (so you can get this comb filter effect to stay on one place.)

I think KGs method is the correct one, that is, it takes the last sample of the ramp inside the saw, and stores it, and the ramp starts from that value when the ramp is active again. The problem is that when I press the same key the memory thing gets bugged after a while (even though both floats have value and are NOT zero, so it should be working but it doesnt). Since i dont know how his memory thing works i cant really fix it.

Exo your method is interesting too, however when the oscillators have drifted apart, and i quickly set their frequency to be the same, and caught a comb filter effect, and press the same key again, it seems the phase starts from a different value. I can only suspect that the blue mono ramp counts differently or something.

Here is my video:
https://www.youtube.com/watch?v=5aIKvPVjUEA

[Xtinct]

Exo your method is interesting too, however when the oscillators have drifted apart, and i quickly set their frequency to be the same, and caught a comb filter effect, and press the same key again, it seems the phase starts from a different value. I can only suspect that the blue mono ramp counts differently or something.

Surely with a free running osc unless the time between key presses is a multiple of the frequency it should be a different phase value.

Edit Ah should have watched the vid was thinking of a different scenario.

[Exo]

Thats good exo, although it still behaves strangely.

I have made a video to show you guys the problems. To me the standard/perfect free running phase is like in many synths like Massive, Rekator 5 etc... they are all doing the same thing so to me thats the correct way.

I demonstrated this using MonoPoly which uses the same method. I have two sawtooth waveforms slightly detuned. Keep an eye on the FFT spectrum, and one can see the phasing effect. There are two things to notice: the phase starts and resumes when the key is let go and pressed again (so we now know it only counts once the osc are active again, its not counting in the background), and the second thing is that, when the osc are detuned, and when i set them the same frequency, they will be drifted apart since the phase difference, however the phase difference stays the same. (so you can get this comb filter effect to stay on one place.)

I think KGs method is the correct one, that is, it takes the last sample of the ramp inside the saw, and stores it, and the ramp starts from that value when the ramp is active again. The problem is that when I press the same key the memory thing gets bugged after a while (even though both floats have value and are NOT zero, so it should be working but it doesnt). Since i dont know how his memory thing works i cant really fix it.

Exo your method is interesting too, however when the oscillators have drifted apart, and i quickly set their frequency to be the same, and caught a comb filter effect, and press the same key again, it seems the phase starts from a different value. I can only suspect that the blue mono ramp counts differently or something.

Here is my video:
https://www.youtube.com/watch?v=5aIKvPVjUEA

Thanks for the feedback and the video helps a lot . I agree getting it to sound the same as on the big synths is what we should be aiming for. You are right KG_is_back version is how it should sound, but what do you mean by the memory thing gets bugged? In my tests when playing about with detuning and playing the same note sometimes the pitch of the note changes every other press like hitting two separate notes is this what you are experiencing?

[adamszabo]

You are right KG_is_back version is how it should sound, but what do you mean by the memory thing gets bugged? In my tests when playing about with detuning and playing the same note sometimes the pitch of the note changes every other press like hitting two separate notes is this what you are experiencing?

Its better to do a video thank try to explain

https://www.youtube.com/watch?v=uUtvw7tdEMw

It seems when you press several keys the memory address for the phase gets broken, thus defeating the whole purpose of poly phase

[KG_is_back]

You are right KG_is_back version is how it should sound, but what do you mean by the memory thing gets bugged? In my tests when playing about with detuning and playing the same note sometimes the pitch of the note changes every other press like hitting two separate notes is this what you are experiencing?

Its better to do a video thank try to explain

https://www.youtube.com/watch?v=uUtvw7tdEMw

It seems when you press several keys the memory address for the phase gets broken, thus defeating the whole purpose of poly phase

I'll look into it by the end of the month... I have exams at the moment, so no time for FS

[tester]

Good luck with exams!

[Exo]

You are right KG_is_back version is how it should sound, but what do you mean by the memory thing gets bugged? In my tests when playing about with detuning and playing the same note sometimes the pitch of the note changes every other press like hitting two separate notes is this what you are experiencing?

Its better to do a video thank try to explain

https://www.youtube.com/watch?v=uUtvw7tdEMw

It seems when you press several keys the memory address for the phase gets broken, thus defeating the whole purpose of poly phase

Thanks for the video, hearing and seeing it is a lot more helpful .


I think the problem is not to do with the memory address hack . I do not fully understand the asm so I am struggling to workout if it is just an error in the code or the problem is inherent in reliance on a circular buffer. But I'm thinking that a circular buffer is not adequate for this and not how other synths are doing it.

From what I have read each voices phase starts from the last phase from the last note played on that voice.
So really we need a way to identify each voice and save the phase for that voice in a consistent index of an array(mem) . Then it's just a case of reading back the phase from the index that corresponds to the voice being played.

I have come up with a very simple and cheap method of tagging voices so we can identify them, just a simple index number that is used directly to read from the mem array.
So voice 1 writes to the first index, voice 2 to the second ect..
Then obviously read back in that order too.

The asm for this should be fairly simple but I need to brush up on it first. I'm pretty sure this method should sound right.

Will be able to finish this on the weekend hopefully .

[Exo]

OK just uploaded the latest prototype, get it in my first post!

I think this is right now so please give this a good test adamszabo

A big thanks goes to Trog for his latest Shared Mem module. I have modified it to work with poly and in a way that makes sense for the poly stream.

[trogluddite]

On first blush, that sounds bang on to my ears, Exo, bravo!

A big thanks goes to Trog for his latest Shared Mem module. I have modified it to work with poly

You are most welcome - what a buzz to see my first schematic after such a long break finding a perfect home less than 24 hours after coming back!
Ingenious work on the poly conversion, too. I don't work with poly much, as I'm more of an FX guy than synths - so I learned some cool things about the workings of poly just from a quick peek at the code.

Off topic.
I don't usually go in for "synchronicity" and all that jazz - but I come back to the forum after 6 months, and I find Exo is back, the stuff I've been working on perfectly dove-tails with some cool forum projects that let us do "impossible" things; and an e-mail from Malc about cleaning up the spam only a few days before. What are the odds?

[Exo]

On first blush, that sounds bang on to my ears, Exo, bravo!

A big thanks goes to Trog for his latest Shared Mem module. I have modified it to work with poly

You are most welcome - what a buzz to see my first schematic after such a long break finding a perfect home less than 24 hours after coming back!
Ingenious work on the poly conversion, too. I don't work with poly much, as I'm more of an FX guy than synths - so I learned some cool things about the workings of poly just from a quick peek at the code.

Off topic.
I don't usually go in for "synchronicity" and all that jazz - but I come back to the forum after 6 months, and I find Exo is back, the stuff I've been working on perfectly dove-tails with some cool forum projects that let us do "impossible" things; and an e-mail from Malc about cleaning up the spam only a few days before. What are the odds?

Thanks, Yes the stars seem to be aligning!

What perfect timing as soon as I downloaded your shared Mem I thought "Yes this is exactly what I need!" 20mins later we now have free running oscs in Flowstone! Unless adamszabo comes back and shows us it's not working right