Cytomic "The Scream" stomp box distortion plugin

[Funkybot's Evil Twin]

Hi Andy, great news on the updates. Glad all that work paid off and we're approaching an official 1.0 version of The Scream. Will there be another public beta beforehand we can look forward to?

I hope you wouldn't mind a few questions on Oversampling in The Scream. Why separate Up/Downsampling filter types (other than cause you can)? Is there ever a use-case where I'd want Minimum Phase upsampling and Linear Phase downsampling or vice-versa? Also, is pre-ringing a concern with Linear Phase oversampling when it comes to The Scream? I've been sticking with Minimum Phase for both, but honestly, have no idea why other than I associate Linear Phase with pre-ringing artifacts and would otherwise like to do as little harm as otherwise possible to the phase response.

I try not to overthink these options too much, but if I'm being honest, I don't fully understand the pros and cons of each approach or why I should be picking one type over another. So a little more information in this regard in the 1.0 manual would be much appreciated. Even if it's just a few bullet points on each like, "Linear Phase: longer latency, pre-ringing, no changes to phase. Minimum Phase: lower latency, no pre-ringing, no changes to phase, etc." Or maybe just include a "Recommended" setting that will work well for all-round usage.

Finally, quick installer suggestion: the installer asks you to select your VST 32-bit and VST 64-bit folders before you even get pick to what formats you want to install. I was thinking, "wait, I don't want to install the 32-bit version why do I have to pick a location" then two screens later was the option to select the plugin types that would be installed. If possible, it would make more sense to swap the order. Have the users select the types they want to install first, then only ask for the relevant folder locations afterwards.

Thanks!

[tristan-]

When things get great there's no point remaining a one man show. There are ways to keep code consistent and quality assured.

[Delta Sign]

I just wanna say this thing is beautiful in combination with ABL3. That's all

[andy-cytomic]

Hi Andy, great news on the updates. Glad all that work paid off and we're approaching an official 1.0 version of The Scream. Will there be another public beta beforehand we can look forward to?

Yep, once I've got it in feature complete form I'll put out a public beta, but I want to tidy up a few things before I do that, and implement the GUI panel for the randomisation engine.

I hope you wouldn't mind a few questions on Oversampling in The Scream. Why separate Up/Downsampling filter types (other than cause you can)? Is there ever a use-case where I'd want Minimum Phase upsampling and Linear Phase downsampling or vice-versa? Also, is pre-ringing a concern with Linear Phase oversampling when it comes to The Scream? I've been sticking with Minimum Phase for both, but honestly, have no idea why other than I associate Linear Phase with pre-ringing artifacts and would otherwise like to do as little harm as otherwise possible to the phase response.

I try not to overthink these options too much, but if I'm being honest, I don't fully understand the pros and cons of each approach or why I should be picking one type over another. So a little more information in this regard in the 1.0 manual would be much appreciated. Even if it's just a few bullet points on each like, "Linear Phase: longer latency, pre-ringing, no changes to phase. Minimum Phase: lower latency, no pre-ringing, no changes to phase, etc." Or maybe just include a "Recommended" setting that will work well for all-round usage.

I added the oversampling phase modes to tradeoff latency vs phase accuracy, and also to be able to match as closely as possible the sound card's bandlimiting filters so you can match ITB what it sounds like to process through external hardware. I needed this feature to accurately match the sound of The Scream to the sound of processing through the TS-808 with my Fireface UCX, it uses an intermediate phase filter on it's output, but a linear phase filter on it's input.

Minimum phase has the lowest latency possible, each frequency is delayed only as much as needed to get the resultant low pass filtering, which is why it's called "minimum". It actually has maximum alteration of the phase in terms of the lower frequencies have a different group delay compared to the higher frequencies. Linear phase delays all frequencies by the same amount. Intermediate phase is in between the two.




I recommend using linear phase for both up and down for most tasks, but if you want low latency for use live, then use minimum phase for both up and down, but realise that repeated processing with minimum phase filters will push the high frequencies further and further back and make things sound a weird. This won't matter if your base rate is 88.2 / 96 kHz, only if you're starting from 44.1 / 48 kHz.

Finally, quick installer suggestion: the installer asks you to select your VST 32-bit and VST 64-bit folders before you even get pick to what formats you want to install. I was thinking, "wait, I don't want to install the 32-bit version why do I have to pick a location" then two screens later was the option to select the plugin types that would be installed. If possible, it would make more sense to swap the order. Have the users select the types they want to install first, then only ask for the relevant folder locations afterwards.

Thanks!

I'm using Inno Setup to create the windows installers ( http://www.jrsoftware.org/isinfo.php ), so I'll have a look if this is possible to first ask for options then ask for user folder locations

[perfumer]

And only for the fomat(s) the user wants to install, others paths remain hidden.

+1 I've noticed that on installers for other apps too. It's very annoying. Some install AAX and that other thing that's not VST without even asking whether you want them or not. Why?

[sleepcircle]

Thank you for your very thorough explanation on the issue! So if I understand correctly, pre-ringing is not a problem in the Linear-Phase option? Like this isn't a FFT system or anything.

[andy-cytomic]

Thank you for your very thorough explanation on the issue! So if I understand correctly, pre-ringing is not a problem in the Linear-Phase option? Like this isn't a FFT system or anything.

I'm not sure what you mean here. Please have a look at the step responses for what the time domain signal looks like, and look at the frequency responses for what will happen to the spectrum, there isn't anything I can say that isn't already shown there!

I can't really make a call on what is a problem or not until I know the context of what you're talking about and what the desired result it. So if you have specific situation in mind you want to ask my opinion about then please go ahead.

[sleepcircle]

No yeah it really doesn't look like a problem at all, hahah--thank you for clarifying everything!

[GrabtharsHammer]

I've just read 50+ pages of tech talk and begin to wonder if, as a musician, I really still belong to the target group of the product.

[sleepcircle]

of course you belong! musicians don't need to know how the hardware is made to be able to use it.

thus:

• you don't need to know about circuit modding (there will be mod presets, and the original distortion pedal is great enough that you don't really need to change it)
• you don't need to know about coding techniques (it'll already have been done by the time you download it), and
• you don't need to know about phase (just make them both linear and call it a day).

proceed to shred lixx

[andy-cytomic]

I've just read 50+ pages of tech talk and begin to wonder if, as a musician, I really still belong to the target group of the product.

You're quire right actually, no musicians allowed, only university graduates in maths and physics are allowed to purchase - I insist upon a scan of your degree at the checkout along with your card details

I'm a fairly techie type of person, and I like to keep people up to date with what is going on, so sorry if the tech talk is all a bit much! Hopefully at least the easy takeaway is that I sweat the details so you won't have to

[acousticglue]

It's like anything else twist the knobs, change parameters and what does it sound like to you?

[Robert Randolph]

I added the oversampling phase modes to tradeoff latency vs phase accuracy, and also to be able to match as closely as possible the sound card's bandlimiting filters so you can match ITB what it sounds like to process through external hardware. I needed this feature to accurately match the sound of The Scream to the sound of processing through the TS-808 with my Fireface UCX, it uses an intermediate phase filter on it's output, but a linear phase filter on it's input.

Minimum phase has the lowest latency possible, each frequency is delayed only as much as needed to get the resultant low pass filtering, which is why it's called "minimum". It actually has maximum alteration of the phase in terms of the lower frequencies have a different group delay compared to the higher frequencies. Linear phase delays all frequencies by the same amount. Intermediate phase is in between the two.




I recommend using linear phase for both up and down for most tasks, but if you want low latency for use live, then use minimum phase for both up and down, but realise that repeated processing with minimum phase filters will push the high frequencies further and further back and make things sound a weird. This won't matter if your base rate is 88.2 / 96 kHz, only if you're starting from 44.1 / 48 kHz.

Andy, in your graphs there, I assume the plots are shifted for readability? Or perhaps the inputs?

Or does your linear phase oversampling filter actually shift the output by 1ms?

[andy-cytomic]

Andy, in your graphs there, I assume the plots are shifted for readability? Or perhaps the inputs?

Or does your linear phase oversampling filter actually shift the output by 1ms?

This is the real deal of what is going on, Cytomic linear phase oversampling delays the output by 1mS, which you can see in the group delay plot - this is the number of milliseconds delay at each frequency. For linear phase the group delay is a constant value, a horizontal line. For other types of filters the delay is non-linear with frequency. This is what linear phase and minimum phase actually mean. I report to the DAW an integer number of samples of latency that the oversampling adds, and the DAW will delay EVERY other channel / bus to at least this number of samples so all audio is lined up again. This is what latency compensation is!

It's easy enough to avoid all this, just run your daw at 88.2 / 96 khz and be done with it. Double all your soundcard buffer sizes and you'll have the same soundcard buffer latency as at 44.1 / 48 khz, but you will have automatically x2 oversampled your entire DAW and all plugins you run, and if you do want to oversample more from there you have 1/2 the latency in milliseconds you would otherwise have had since your sample rate is twice as fast.

[Robert Randolph]

Andy, in your graphs there, I assume the plots are shifted for readability? Or perhaps the inputs?

Or does your linear phase oversampling filter actually shift the output by 1ms?

This is the real deal of what is going on, Cytomic linear phase oversampling delays the output by 1mS, which you can see in the group delay plot - this is the number of milliseconds delay at each frequency. For linear phase the group delay is a constant value, a horizontal line. For other types of filters the delay is non-linear with frequency. This is what linear phase and minimum phase actually mean. I report to the DAW an integer number of samples of latency that the oversampling adds, and the DAW will delay EVERY other channel / bus to at least this number of samples so all audio is lined up again. This is what latency compensation is!

It's easy enough to avoid all this, just run your daw at 88.2 / 96 khz and be done with it. Double all your soundcard buffer sizes and you'll have the same soundcard buffer latency as at 44.1 / 48 khz, but you will have automatically x2 oversampled your entire DAW and all plugins you run, and if you do want to oversample more from there you have 1/2 the latency in milliseconds you would otherwise have had since your sample rate is twice as fast.

Right, I understand the implications of the filters and PDC. I was curious about the implementation in terms of output in a DAW.

Is the 1ms PDC compensated?