Cytomic "The Scream" stomp box distortion plugin

[andy-cytomic]

An A/B with some guitar loops:
https://cytomic.com/files/scream-vs-ts8 ... ive050.mp3

The difference I can hear is in the lower mids. There's more bloom in the A examples, while B produces a more controlled overdrive.

Overall, you've nailed the general overdrive character, but some tuning has to be done to the frequency distribution of that drive.

Can you please be more precise with what you exactly mean by "bloom"? Which range of frequencies are you talking about in hz? Also can you please let me know which time in the audio your are talking about that best demonstrates this? Also which time in the test signals best exhibits this?

From what I can hear the majority of the difference is due to noise in the analog circuit. I've only matched a handful of the components so there are bound to be differences in the tone. I can also post audio from an analog TS9, which sounds a lot different to the TS808 I've got - but they are exactly the same circuit apart from two resistors in the output section! The majority of the difference comes from the exact value and types of the diodes in the distortion section, and then the variation of the potentiometer values which can be as much as 20% off from marked, then very subtle differences from the exact values of each of the components and ics.

For a laugh I could post some of the competitor's models which don't really come close to matching huge chunks of the character, but that would be mean

edit: I just realised although I desoldered and measured the resistance of the drive knob to get the taper right, I forgot to actually implement this!! I'lll repost the example with the correct amount of drive between the two examples. I mostly have been testing either at full drive or zero drive, not at 50% which is where I posted these examples of the guitar

[izonin]

edit: I just realised although I desoldered and measured the resistance of the drive knob to get the taper right, I forgot to actually implement this!! I'lll repost the example with the correct amount of drive between the two examples. I mostly have been testing either at full drive or zero drive, not at 50% which is where I posted these examples of the guitar

This might be it. The A examples seem to have more drive than B.

[andy-cytomic]

edit: I just realised although I desoldered and measured the resistance of the drive knob to get the taper right, I forgot to actually implement this!! I'lll repost the example with the correct amount of drive between the two examples. I mostly have been testing either at full drive or zero drive, not at 50% which is where I posted these examples of the guitar

This might be it. The A examples seem to have more drive than B.

Yes, a bit more drive and also the 1 pole low pass post drive was set a little too high. I've matched things a bit more carefully now, thanks for the feedback

https://cytomic.com/files/scream-vs-ts8 ... ive050.mp3

[MFXxx]

And, with the permission of the acidvoice guys (http://www.acidvoice.com), a TB-303 saw pattern:
analog ts808 - https://cytomic.com/files/scream-acidvo ... -ts808.mp3
scream - https://cytomic.com/files/scream-acidvo ... scream.mp3
scream with mods - https://cytomic.com/files/scream-acidvo ... am-mod.mp3

The acidvoice guys sent me .wav files to process, but I'm still waiting for the a version of the pattern they use on the distortion page, will post when it arrives.

Dam close, there's a slight variation but this is very slight to my hearing [around 36-39secs] and the sweep stepping sounds slightly different, but maybe this is too be expected.

Sounding very very good Andy, Cant wait.

[meloco_go]

Yes, a bit more drive and also the 1 pole low pass post drive was set a little too high. I've matched things a bit more carefully now, thanks for the feedback

https://cytomic.com/files/scream-vs-ts8 ... ive050.mp3

This is impressive, really close!

[izonin]

https://cytomic.com/files/scream-vs-ts8 ... ive050.mp3

B still sounds a bit rounder, but the differences are now tiny. I can mainly hear it in the 100-200Hz range where B has a bit more weight.

[andy-cytomic]

https://cytomic.com/files/scream-vs-ts8 ... ive050.mp3

B still sounds a bit rounder, but the differences are now tiny. I can mainly hear it in the 100-200Hz range where B has a bit more weight.

Thanks for the continued excellent feedback! I agree that B is still smoother and better sounding, and I was really puzzled because clearly in the sin / noise tests there is no discernible difference.

I tried out a square wave to see how the DC, but the DC looked fine. The square wave did, however, show me what the culprit was. I could see a strange little ramp that wasn't in the analog version that introduced some mid to high mid harshness to the sound. Here is an image so show what is going on.

key:
red=final output
blue=input
lightblue=highpass that gets clipped by diodes
green=output of drive op-amp
lightbrown=output of eq op-amp

Linear phase with problem little bump in the final output:


Minimum phase, problem gone:


The pre-ringing of the linear phase oversampling filter was getting clipped off one side by the diodes, then the other side of the wobbles gets low pass filtered and makes a little ramp where there shouldn't be one. It may look small, but it's big enough to cause an issue in the smoothness of the sound.

I'll have to implement minimum phase oversampling as an option and set this to the default for The Scream, so this will probably delay the release a bit. I'll post a new example once it's done, hopefully in the next couple of days.

(edit: this would happen in analog as well if the stomp box was sent a linear phase filtered signal, but my sound card has a minimum phase bandlimiting filter)

[meloco_go]

This is really interesting!
I wonder if such small things are the culprits in many situations when people claim there is a difference yet usual tests show there's no.

[bmanic]

It depends on what is affected. The only reliable way to know if the difference is large enough to be of concern is to do a proper ABX test.

Sometimes even if the difference signal is down to -90dBFS you can still correctly find the desired process in an ABX test. Usually this comes down to transients.. as soon as the transient is even a tiny bit altered, a critical listener will hear a difference. It's much more difficult on steady state stuff and when the waveform is steady and complex (aka noise).

I've done these kinds of ridiculous shootouts on my own compressor designs and always been surprised how sensitive the ear can be to tiny changes. I've also surprised myself the opposite way.. when the difference signal is quite high yet I heard no difference.

This is what makes critical listening so important and not just looking at test tones and mathematical results.

[andy-cytomic]

I've added optional minimum phase upsampling for the first stage to the oversampling engine so what signal my model gets to process matches more closely what the soundcard outputs, and hence what the ts-808 gets to process. The model hasn't changed, but now a bit of the mid to upper mid harshness in the A signal is removed to it is very close to the B signal, apart from hiss:
https://cytomic.com/files/scream-vs-ts8 ... ive050.mp3

And for refence here is a ts808 and a ts9:
https://cytomic.com/files/ts808-vs-ts9- ... ive050.mp3

[andy-cytomic]

This is really interesting!
I wonder if such small things are the culprits in many situations when people claim there is a difference yet usual tests show there's no.

This is a pretty specific issue due to the fact a high passed signal is being boosted x118 (+40 dB), and then hard-clipped to make it asymmetric, but yes if the circuit and the model get different audio then there will be differences!

[meloco_go]

This is really interesting!
I wonder if such small things are the culprits in many situations when people claim there is a difference yet usual tests show there's no.

This is a pretty specific issue due to the fact a high passed signal is being boosted x118 (+40 dB), and then hard-clipped to make it asymmetric, but yes if the circuit and the model get different audio then there will be differences!

I was thinking more in line of spectral difference vs waveform difference. Clearly such things would be more obvious on transients.
I wonder if, say, compressor detector faces such a waveform, shouldn't there be case when some sort of "look ahead" action happens?

[Funkybot's Evil Twin]

I've added optional minimum phase upsampling for the first stage to the oversampling engine so what signal my model gets to process matches more closely what the soundcard outputs, and hence what the ts-808 gets to process. The model hasn't changed, but now a bit of the mid to upper mid harshness in the A signal is removed to it is very close to the B signal, apart from hiss:
https://cytomic.com/files/scream-vs-ts8 ... ive050.mp3

Sounds darn impressive. I'm certainly not hearing a difference. At least nothing I'd be able to identify in any kind of honest testing.

[andy-cytomic]

I was thinking more in line of spectral difference vs waveform difference. Clearly such things would be more obvious on transients.
I wonder if, say, compressor detector faces such a waveform, shouldn't there be case when some sort of "look ahead" action happens?

Setting the threshold increases the gain of the input signal being sent to the detector, so yes, having some pre-ringing could activate the compressor slightly before the main transient. The thing with compressors though is that the operate on the absolute value of the input, so you would at least get symmetric behaviour, unlike the example with The Scream and TS-808.

The main point to take away here is that exactly the same thing will happen with both a good analog model and the actual analog unit being modelled if they are sent the same input signal with pre-ringing. This is possibly a warning to keep an eye (ear?) out for such things when doing any processing, analog or digital, on audio that has been stored digitally and so band-limited when a lot of gain combined with large non-linearities.

[andy-cytomic]

Sounds darn impressive. I'm certainly not hearing a difference. At least nothing I'd be able to identify in any kind of honest testing.

Great, thanks for the feedback!