Asking hardware synth owners for a favor :)

[aciddose]

Does voice recycling mean that the voices on the same note are consecutive rather than partially in parallel?

Recycling means the same voice is used for the same note.

For example in a piano if you hit the same key over and over, the same "voice" (strings) are used over and over. The "envelope" is just "retriggered" by the hammer.

As an example of "severe voice limitation" I'm frequently pissed off that I can only get ~75 voices in Xhip because I'm trying to do a 16-poly 5x unison in real-time. 16 X 5 = 80.

Dropping the unison down changes the sound, although 3x might be "okay" in combination with a chorus effect... sometimes.

Dropping the polyphony is a big deal, voices are stolen like crazy at 12, at 8 it's just a nuthouse of voice stealing and sounds like total shit for a pad.

So I consider 12 or less than 80 "severe".

In a poly-synth like a JX- or Juno where you have six voices? The voice stealing is near intolerable for anything with a long release.

[fluffy_little_something]

Does voice recycling mean that the voices on the same note are consecutive rather than partially in parallel?

Recycling means the same voice is used for the same note.

For example in a piano if you hit the same key over and over, the same "voice" (strings) are used over and over. The "envelope" is just "retriggered" by the hammer.

I see. And does that mean that it is a full restart or is there some carryover? Even with a piano hitting an already vibrating string again might influence the way the same note sounds the second time around. At least that is what might happen in the case of a guitar, I suppose, especially if the note is played very lightly the second time, i.e. the finger/pick does not completely stop the vibration before plucking the string again.

[aciddose]

There must be "carryover" or it will introduce clicks. For example the envelope starts from the position it was already at, not from zero.

It works the same as a monophonic synthesizer would, only like N monophonic synthesizers.

Without recycling new voices are generally reset to zero. The "extra" voices are used when no other voice is available, and the old voice is often forced to fade out quickly, although not always.

As a designer you would recognize the many variations possible.

For example recycling can be combined with "greedy allocation" such that voices are only recycled when no "fresh" voice is available.

[fluffy_little_something]

Is there no init patch in Xhip?
Anyway, judging from the Rich Strings preset it sounds the way it does in Noizemaker, i.e. the way I think it should sound. Neither is anything hacked off, nor is there an unpleasant overlapping and change of the sound character. Well done

[aciddose]

You can force it to do that if you use the alpha and uncheck "recycle" on the bottom left.

The init preset is loaded by hitting the "Clear" button.

[Ingonator]

Just found this concerning my Korg DW-8000 (has digital oscs and an analog filter):

People always think the architecture is identical inside both DW and DSS. It's not apparently! The DSS is a sampler, the DW is not. In fact, I've found out that the DW's waveforms are constantly cycling even when notes aren't being played just like in a real analog oscillator.

This gives more movement to the sound since a key depression will start each wave at different periods, whereas the DSS will start the wave at the same place every time. In fact, in the DW's case there's also a slight difference in phase between whole voices with each one sounding a tiny bit different than the others. So it acts much more as a real analog in this regard...

Overall while playing it indeed seems to be closer to the behavior of analog oscillators than a typical sample playback. The standard waveforms in combination with the filter could sound really fat and sometimes almost give the impression of VCOs. The Osc detuning also sounds quite good and/or "clean", no real "phasing" like with some other synths, same when playing chords.

A lot of plugins seem to include an option to use free running oscillators. The Tone2 Icarus Bete a i am currently testing has this too (with the "Phase" knob and also Waldorf Largo, just to mention 2 examples that are more on the "digital" side. In Largo using free running Oscs seems to be especially important when using the Unison. In Largo if you set the Osc Phase to a fixed value you could get a phasing sound at the attack which later seems to disappear when holding the note. With setting the phase to "free" the attack sems to be clean from phasing.

In Waves Element 2 you got the DCO/VCO swicth in the Oscs where VCO corresponds to free running Oscs. With both set to DCO you could get phasing also depending on the Fine tune amount (more serious with small amounts).

[aciddose]

It actually doesn't make much difference if the oscillators phase is allowed to idle or not.

What matters more is whether LFO (modulator) and envelopes are allowed to idle.

In Xhip this happens, in many other synths it doesn't.

Another huge mistake in my opinion if not present or not an option.

[fluffy_little_something]

Ah, found the Clear button, it so dark I did not see it

Regarding the topic, funny that two free synths do it right, while Sylenth1 sucks in this respect. But according to the developer he wanted it to sound the way it does (which is the 1st half of the first audio file and the 2nd half of the second audio file. The 2nd half of the first audio file is Ultra Analog 2, the 1st half of the second audio file is Zebra.)

[fluffy_little_something]

Ingo, it does not have to do with osc phase as switching phase retrigger on and off has zero impact on the effect this topic is about. I have tried it. It has to do with the overlapping of two or more voices on the same note and the resulting effect, if I am not mistaken.

[Ingonator]

Ingo, it does not have to do with osc phase as switching phase retrigger on and off has zero impact on the effect this topic is about. I have tried it. It has to do with the overlapping of two or more voices on the same note and the resulting effect, if I am not mistaken.

I found that with certain synths like e.g. Largo the Osc Phase is indeed important. Especially when using the Unison with detuning you could get phasing at the attack when setting teh OSc phase to a fixed value. Same in Waves Element 2 when using teh DCO mode for both Oscs. Those are just 2 examples.

Anyway so far i could not find unwanted issues with my Korg DW-8000 (which uses 16 digital DWGS waveforms but seems to use some kind of free running oscs), no matter if it is about Osc phase, envelopes, overlapping voices (DW-8000 has 8 voices with an analog filter chip for each voice), or Osc detuning. Only possible problem is clicking when Amp envelope attack is at 0 but no real problem at higher values.

[aciddose]

The reason osc phase idling isn't too important is that you have to consider "what stops the phase from progressing?"

The note-off does. So it depends upon the sustain, which depends upon the input.

This is quite close to being random, so the start-phase you get when a voice is turned on freshly is nearly random anyway, making it very difficult if not impossible for a human to notice any difference.

It matters a bit in cases like unison where you have several voices starting and stopping at exactly the same time. In which case the phasing effect of the unison will be at the same exact position when the note is turned on as when it turned off.

That said, there is a significant period during the lowest portion of the release cycle of the envelope during which the phase still advances although the voice is otherwise inaudible, especially in combination with other active voices.

In those cases we get an additional "push" forward which makes the effect far more difficult to recognize.

That said "phase idling" is as easy as adding: (phase + idle_time) each block, which is extremely cheap. At the end of the silence (when voice becomes active) phase = wrap(phase), which is again extremely cheap.

The exact same process is used to idle LFOs (merely slower oscillators) and envelopes. In the envelope a slightly more expensive computation is used involving log() where we compute the level of the envelope given some amount of time having passed.

per block: envelope.idle_time += idle_time

on start: envelope.level = compute_decay(envelope.level, envelope.idle_time)

Where idling really matters is for tempo-sync'd LFOs and the like. Try a very slow PWM sound with immediate attack/release time. Hit the same note over and over. You should hear the LFO slowly oscillating at the specified rate such as 5 seconds.

If the phase isn't "free running" it won't run while the note is off, which means it will be much longer than 5 seconds depending upon the ratio between on/off time.

[fluffy_little_something]

I found that with certain synths like e.g. Largo the Osc Phase is indeed important. Especially when using the Unison with detuning you could get phasing at the attack when setting teh OSc phase to a fixed value. Same in Waves Element 2 when using teh DCO mode for both Oscs. Those are just 2 examples.

I remember that phasing effect from my cheap SuperWave Professional and Trance Pro synths when switching to 7 waves per osc in a given instance of the synth. After launching an instance I had to play around for like 15 secs until the 7 waves were finally out of phase. Before that there was that ugly bump at the beginning of each note, followed by a phasing effect. Once that instance had "warmed up", I could switch back and forth between single-wave and multiple-wave modes and between patches, the phasing problem would not reappear. It posed a problem when mixing down as my DAW starts every synth anew before exporting. So the first 15 secs would sound pretty ugly

[aciddose]

https://soundcloud.com/xhip/pwnpwm

Here is the LFO thing I mentioned.

If you play this (mono) with a synthesizer where the LFO phase isn't running during idle you should notice the huge difference right away.

[Ingonator]

I found that with certain synths like e.g. Largo the Osc Phase is indeed important. Especially when using the Unison with detuning you could get phasing at the attack when setting teh OSc phase to a fixed value. Same in Waves Element 2 when using teh DCO mode for both Oscs. Those are just 2 examples.

I remember that phasing effect from my cheap SuperWave Professional and Trance Pro synths when switching to 7 waves per osc in a given instance of the synth. After launching an instance I had to play around for like 15 secs until the 7 waves were finally out of phase. Before that there was that ugly bump at the beginning of each note, followed by a phasing effect. Once that instance had "warmed up", I could switch back and forth between single-wave and multiple-wave modes and between patches, the phasing problem would not reappear. It posed a problem when mixing down as my DAW starts every synth anew before exporting. So the first 15 secs would sound pretty ugly

Recently i had an interesting experience while testing the upcoming Tone2 Icarus. With the Hypersaw Unison modes (in Icarus they work with all waveforms, not just a Saw) i noticed a seriopus phasing effect when detuning the waveforms (up to 9 detuned waves) to lower up to medium amounts (higher amounts seemed to be OK). I did not remeber that from a real Virus TI or samples i had done from it before selling it and also not with the Hypersaw modes in Electra 2.1.
First i got the reply that this could not be replicated and i was close to thinking i was just "hallucinating".
After sending an audio demo it turned out that this was only found with the 64-bit VST2 plugin for Windows but not with 32-bit plugin which then i was able to confirm. Anyway it was finally solved and i am very happy about that.

[fluffy_little_something]

That's why you are a beta-tester, mission accomplished