Whats the deal about snappy envelopes? What about Renoise?

I read about snappy envelopes a lot. Especially when somebody talks about Sylenth 1 or some classic digital /analog synths that are great for percussion.
But i dont get it. Why Sylenth1 should be any faster than any other VSTi synth (that has logarithmic or linear envelopes)?
And what about Renoise, are its envelopes fast enough? And are they snappy or clicky (https://www.muffwiggler.com/forum/viewtopic.php?t=53372) :D?

Or is it a bullshit at all?

Just curious…

I don’t know for sure, but maybe some hardware or software synth vendors will interpolate fast volume changes in a lame way or prevent too fast attacks to prevent implementing volume interpolation. A lot of what’s being discussed by the masters of mastering tends to be voodoo, in my opinion. But transients overall have a great influence on the song and also are important for proper depth localization aka the less the attack the more in the background.

Sylenth sounds ok, but also not special or incredibly good. It also only has standard features. No sync or ring. I believe its still some hype for trance producers, because there is an 303 acid bass line preset in the default bank. :slight_smile:

For me a good synth will only interpolate extreme ramps over some small amount of samples and let the user choose what curve it should have (like linear, exponential, logarithmic etc). I think all linplug (and also papen) synths have here the most flexible options. Korg (and for sure yamaha, roland etc too) of course, too, in a more static way like: log hard/mid/soft, linear and exponential. I don’t have sylenth, but as far as I can remember, it has absolutely no curve options, so I guess it’s linear.

I was going to say that renoise envelopes are a bit too slow for that and don’t have the resolution of 1ms as it would say on the sample modulation devices, but I’ve tested them right now with a sine wave and it is, soI’m a little confused here that’s not how it was in beta

Knowing a bit about dsp coding it does make differences at what rate and style such stuff is interpolated. That also goes for filter cutoffs etc. The measure is probably 100% analog synths where time resolution is close to infinite. In software one can choose an interval at which to update parameters, i.e. in blocks of samples, on sample level, or even deeper with supersampling or integration of change curves. When a synth makes clicky noises, its modulation sucks, maybe often synths will just interpolate/ramp values between update blocks to counteract. Per sample, supersampling or integrative math is of course expensive to the cpu.

The difference won’t really be heard for “slow” changes in values, but when it goes to very hard pitch ramps, volume ramps, filter cutoffs, well - then crappy interpolation won’t create the same “snappy direct” transients that real analog stuff might be able to create. I think for percussion attacks this makes differences. Also the curvature of the modulation shapes the way these “clicks” will sound like. And I don’t mean the clicks produced by crappy amplitude modulation, but intended ones.

Hm but adsr usually is calculated sample-wise in soft synths. The block size should have no influence on adsr or automation if right coded. I would also say that an interpolated adsr will less click, but be less snappier, if done too simple. If a synth has no interpolation, a steep attack will of course click. The best attack curve would smooth out only the top edge, so the main changing energy will remain. Synthmaster e.g. has some approach like this.

Another argument against snappy analog vs digital might be: usually a classic analog synth uses free oscs, which you cannot restart. So sometimes a bass already will sound kind of mini delayed, if the first oscillation cycle is not a complete cycle. The lower the note the stronger the effect (a bass sound may not need a snappy attack anyway :slight_smile: ) A sample on the contrary will always restart and may sound snappier of course. A free osc is one typical analog sounding feature (I would never make a bass with a fixed osc), which I miss in samplers like renoise a.k.a. sample start offset as modulation target (random modulated and pitch dependant). You can work-around this in r3, using a short, spiky modulating pitch to delay the phase.

@Jurek, you know you could do that with phrasesand with a 0Sxx and 0Yxx comands

@Jurek, you know you could do that with phrasesand with a 0Sxx and 0Yxx comands

Thanks, that’s a good idea, only believe it’s too rough with these commands, or can I find adjust the phase (so only increase start point by some samples)?

@Jurek: If you use a periodic waveform, for a 256 frame long sample that’s per sample :stuck_out_tongue: (resampling aside) It’s not analogue-free-form no, but I don’t see a point in striving to get the old in the best precision possible, the new does the same thing slightly differently, embrace it not as a limitation.

For those wondering, quote doesn’t work in IE, therefore the @

Oh, ok it’s relative to sample length. That’s really some good approach.

Hm but adsr usually is calculated sample-wise in soft synths. The block size should have no influence on adsr or automation if right coded. I would also say that an interpolated adsr will less click, but be less snappier, if done too simple. If a synth has no interpolation, a steep attack will of course click. The best attack curve would smooth out only the top edge, so the main changing energy will remain. Synthmaster e.g. has some approach like this.

Hi there!

SynthMaster actually calculates all LFOs/envelopes every block not every sample. The block length is set by the “engine size” parameter. Small = 16 samples (0.3 ms) , Normal = 32 samples (0.6 ms), Large = 64 samples (1.2 ms)

Hi there!

SynthMaster actually calculates all LFOs/envelopes every block not every sample. The block length is set by the “engine size” parameter. Small = 16 samples (0.3 ms) , Normal = 32 samples (0.6 ms), Large = 64 samples (1.2 ms)

Thank you for clarifying this.

So, between these step widths, it will be kind of spline interpolated dependant on the chosen curve? So would you say, for most snappy attacks you should choose the lowest setting? Or is it more important for tight timings?

No we actually use linear interpolation for sample accurate parameter updates.