Ring Mod Sidechain

[valankar]

Hi all,

Recently I learned about ring mod sidechain, mentioned in these videos:

Decided to do a deeper dive into it in these videos in case anyone interested:

Enjoy.

[Paralytik]

Wow, this is pretty insane. I've never heard about this technique before.

[valankar]

It really has been a game-changer for me. Previously kicks might cause over 0db peaks with normal sidechaining, but with ring mod sidechaining those can peak at -4, -5 db. And it actually doesn't sound bad.

It definitely changes the way I mix though, as previously I would lower those kicks, or apply clipping, etc. So you really need to listen a lot more closely and not rely too much on metering (maybe for the better).

[orthodox]

Having seen that, I just thought of another technique that I can implement with compressors I have. The idea is to ring-modulate the kick with itself or "square" it in order to get uni-polar "rectified" signal and then use it to ring-mod the ducking target.

[selig]

Still the cleanest non distorting side chain I've come up with is cross fader. If the goal is to not raise the peak level while also not distorting the waveform in any way, it's the most accurate/surgical (and foolproof) way to accomplish this.

[valankar]

Hi all,

I did a newer video that is more Reaper focused, but can be applied to other DAWs. Just wanted to showcase ringmod sidechain with something a bit more musical. Hope its useful:

Still the cleanest non distorting side chain I've come up with is cross fader.

When I thought about this more, this is essentially what ringmod sidechain is doing, but at the sample level.

[selig]

Hi all,

I did a newer video that is more Reaper focused, but can be applied to other DAWs. Just wanted to showcase ringmod sidechain with something a bit more musical. Hope its useful:

Still the cleanest non distorting side chain I've come up with is cross fader.

When I thought about this more, this is essentially what ringmod sidechain is doing, but at the sample level.

Not sure I follow - the distortion caused by ring mod would always be audible, no? If you duck a sine wave with ring mod and listen to the results you will hear additional harmonics (distortion) being generated, right? You should hear the kick modulating the bass/sine easily if I’m not mistaken, which is often masked a bit when using complex waveforms.
The examples I’ve see use very bright bass tracks which would mask any distortion, so it’s hard to tell. Do you have a moment (and the motivation) to try RM ducking on a sine/sub bass to hear how much distortion is imparted on the bass signal?

OTOH, using a crossfader means you can only hear one or the other - and even if you DID hear both they would be reduced by 6 dB by the pan law, which is more than enough to compensate for the 6 dB max increase when two sounds play together. Meaning, there is no case where you can INCREASE gain with this setup no matter what you do, which is the goal of ducking, right?And you won’t hear distortion this way either since you’re not using audio rate modulation on the signals - you’re just crossfading between two signals.
So while the crossfade method is not ‘perfect’, it’s the closest thing to perfect that I’m aware of since you automatically achieve your ‘gain control’ goals no matter the setting. Plus, it requires no real calibration except to be sure you send enough CV signal to fully crossfade with each kick which is easy to check if you meter the kick to make sure it’s hitting the same level as before adding this to the equation. Hopefully I’m making sense, and surely someone else must have thought of doing it this way?

[valankar]

Not sure I follow - the distortion caused by ring mod would always be audible, no? If you duck a sine wave with ring mod and listen to the results you will hear additional harmonics (distortion) being generated, right? You should hear the kick modulating the bass/sine easily if I’m not mistaken, which is often masked a bit when using complex waveforms.
The examples I’ve see use very bright bass tracks which would mask any distortion, so it’s hard to tell. Do you have a moment (and the motivation) to try RM ducking on a sine/sub bass to hear how much distortion is imparted on the bass signal?

OTOH, using a crossfader means you can only hear one or the other - and even if you DID hear both they would be reduced by 6 dB by the pan law, which is more than enough to compensate for the 6 dB max increase when two sounds play together. Meaning, there is no case where you can INCREASE gain with this setup no matter what you do, which is the goal of ducking, right?And you won’t hear distortion this way either since you’re not using audio rate modulation on the signals - you’re just crossfading between two signals.
So while the crossfade method is not ‘perfect’, it’s the closest thing to perfect that I’m aware of since you automatically achieve your ‘gain control’ goals no matter the setting. Plus, it requires no real calibration except to be sure you send enough CV signal to fully crossfade with each kick which is easy to check if you meter the kick to make sure it’s hitting the same level as before adding this to the equation. Hopefully I’m making sense, and surely someone else must have thought of doing it this way?

Yes you are right. I was originally thinking ringmod sidechain was like moving a crossfader very fast, but instead it is like a compressor with extremely fast/instant attack and release. There is definitely a not-so-pleasant distortion (with a sine wave, and also anything else), which I guess is expected. It sounds similar to a compressor setup that way.

The crossfader method does seem the better route. I've seen a few videos where people merge kick and bass samples in this way in the timeline, with crossfading in the generated sample. I haven't seen anyone using a plugin. Perhaps in some DAWs it is not sample-accurate to do that with automation. But I could imagine an LFOtool-like device that instead of ducking according to the envelope, it crossfades between the input and sidechain.

[valankar]

Here is a video that is essentially about doing the crossfading method, except they use 2 lfo plugins with inverse shapes:

[selig]

Here is a video that is essentially about doing the crossfading method, except they use 2 lfo plugins with inverse shapes:

I guess the main difference would be you can still ‘miss’ transients if you’re not aggressive enough and precise enough, meaning both can still hit at the same time if you’re not careful with the setup. Whereas with a crossfader it is impossible to EVER have both play at the same time and increase level because even if you are at the 50% point on the crossfader, the crossfade curve ensures the overall level doesn’t increase - like magic!

[valankar]

I did a small video on using the crossfade technique. It's more Reaper specific, but similar with other DAWs. I think I like it better than the other methods, and can even do it with multiband .

[selig]

Great video!
Some notes/thoughts:

The high hat sounds cleaner on the crossfade vs ring mod examples, making it sound softer (but IS it really softer?)

If you control the crossfade with the envelope of the drums you won't likely have the level issue (where you had to lower the level by a tiny bit). I've never seen any "overs" when doing crossfade ducking, but you are likely seeing it because of how you are controlling the crossfading if I'm not mistaken. You want the ONLY duck when drums are hitting.
You can also extend the range of the control signal which pushes the crossfader to it's extremes so it doesn't spend any time in the center positions, making it act more like a nice smooth switch. This is accomplished by what modular synth users call CV Clipping, which produces a result that 'pegs' to either side more of the time with minimal time spent in the center - hopefully that's clear, but maybe not something you can do in Reaper (but is easy to do in Reason any number of ways).

Finally, using two copies of the crossfader (one on each signal instead of both signals into one crossfader as per usual) allows you to keep kick/bass separate (on their own tracks) and still behave exactly the same. In Reason you put the controls on a Combinator so one set of controls affects both devices equally.

Another quick note: the bass sounded lower in the multi-band version to me (around 8 min), but could just be my imagination!

[valankar]

If you control the crossfade with the envelope of the drums you won't likely have the level issue (where you had to lower the level by a tiny bit).

Ah interesting, I didn't think of using the drum envelope! Will need to try that.

You can also extend the range of the control signal which pushes the crossfader to it's extremes so it doesn't spend any time in the center positions, making it act more like a nice smooth switch.

Hm, I'm not sure I get that. Let's say in general I have 1 drum track (with everything, kick, snare, hats, etc). It's only when the drums hit hard with a kick that I want to duck the other tracks. But for the rest of the time, I'm thinking it should mix 50/50. If it's a switch, then in this case I'd just hear the kick and the rest of the drums would be silent once it fades over. Or perhaps you are thinking of some different track setup? I'm thinking there are only 2 buses I can crossfade between: drums, everything not drums.

Another quick note: the bass sounded lower in the multi-band version to me (around 8 min), but could just be my imagination!

When I went back to listen you are right. I suspect the multiband crossover is messing with the phase. I tried the linear phase crossover in MXXX and it is a little better but still kinda off. Maybe it's best to just stick with fullband.

Thanks!

[aquil]

A Bitwig user brought this topic to my attention, and I wanted to try recreating it in Reason. Currently, I'm using the CMD Rack Extension, which allows for mathematical manipulation of audio signals. Ideally, if the RM Sidechain is set correctly, there shouldn't be any peaks over 0dB, assuming both audio signals being processed have a maximum peak of 0dB. However, in my initial attempts, peaks above 0dB kept occurring.

One reason seems to be the parallel audio output in the SSL Mixer track. As soon as I use it to create, for example, a duplicate of the kick signal, phase cancellation no longer works correctly. My second attempt was to duplicate the kick audio signal via the mixer channel's effect rack and send it to the RM Sidechain, but this also doesn't cancel out. The only solution that worked was routing the signal through the Direct Out, then adding the kick audio signal back in at my RM routing using the audio merger tool.

It seems like, despite PDC, Reason doesn't have stable timing for the audio signals, or maybe I've misrouted something.

[joeyluck]

kHs Ring Mod has sidechain and it's free.

https://www.reasonstudios.com/shop/rack ... -ring-mod/

[aquil]

I know, but with the KHS Ring Mod, I can't get the ring mod sidechain to stay below 0dB. Something isn't right there, and it also adds an 7-sample delay. MATH currently has zero sample delay. I'm wondering why any delay is necessary at all with the KHS Ring Mod.

Edit: Hmm, I just tried the KHS Ring Mod again. As expected, it goes over 0dB. But I discovered something interesting: when I go directly from the mixer track where the kick is, via direct out to the sidechain input of the KHS Ring Mod, there's an 7-sample delay. However, if I route it through an audio splitter first and then into the sidechain, the delay is zero samples. This might be a bug—I might submit a bug report.

[Loque]

I know, but with the KHS Ring Mod, I can't get the ring mod sidechain to stay below 0dB. Something isn't right there, and it also adds an 7-sample delay. MATH currently has zero sample delay. I'm wondering why any delay is necessary at all with the KHS Ring Mod.

Edit: Hmm, I just tried the KHS Ring Mod again. As expected, it goes over 0dB. But I discovered something interesting: when I go directly from the mixer track where the kick is, via direct out to the sidechain input of the KHS Ring Mod, there's an 7-sample delay. However, if I route it through an audio splitter first and then into the sidechain, the delay is zero samples. This might be a bug—I might submit a bug report.

AFAIR Reason adds a delay in some situation's to prevent feedback loops with 0ms delay, which would just generates a endless peak sound. The delay time is related to your buffer size.

[ShadowMatriks]

Never heard of this before interesting.

For midi sequenced my preferred way in Reason is to duck using a ghost Kong, Octorex, Redrum, whatever duplicate in a Combinator (so both devices use the same midi notes and samples) with the ghost devices audio output going into Pulveriser. I use the CV follower signal from Pulveriser (which has threshold, attack and decay) to duck bands on the M-Class EQ (setup unipolar in a Combinator. It's response can be trimmed using the controls on Pulveriser, the combi trim and/or editor) usually two bands per kick or snare. I prefer this method to the compressor ducking the whole bass signal, as I only ever do that with four on the floor music which needs the bounce. As far as I can tell there is no attack delay like with a compressor!

You can decide which hits trigger the eq regardless of whether it has it's own audio channel and without interfering with the function of the main output device, by muting the ghost devices channels, removing the surplus samples or using slice edit in Octorex as well as the individual outs! The ducking signal is completely definable, with all the controls of the Pulveriser and combi trim and editor at your disposal. I've found it to be the most flexible, has the least amount of issues (missed hits and delay) and can be used with midi tracks that have lot's of edits and complex patterns.

[selig]

Never heard of this before interesting.

For midi sequenced my preferred way in Reason is to duck using a ghost Kong, Octorex, Redrum, whatever duplicate in a Combinator (so both devices use the same midi notes and samples) with the ghost devices audio output going into Pulveriser. I use the CV follower signal from Pulveriser (which has threshold, attack and decay) to duck bands on the M-Class EQ (setup unipolar in a Combinator. It's response can be trimmed using the controls on Pulveriser, the combi trim and/or editor) usually two bands per kick or snare. I prefer this method to the compressor ducking the whole bass signal, as I only ever do that with four on the floor music which needs the bounce. As far as I can tell there is no attack delay like with a compressor!

You can decide which hits trigger the eq regardless of whether it has it's own audio channel and without interfering with the function of the main output device, by muting the ghost devices channels, removing the surplus samples or using slice edit in Octorex as well as the individual outs! The ducking signal is completely definable, with all the controls of the Pulveriser and combi trim and editor at your disposal. I've found it to be the most flexible, has the least amount of issues (missed hits and delay) and can be used with midi tracks that have lot's of edits and complex patterns.

Pulveriser is quite slow to respond in my testing, even the MClass Compressor is faster. But faster still are Sweeper and the new Sidechain Tool in R13 (which is actually a compressor behind the scenes, as is Sweeper if configured correctly).I guess that's what you are calling "attack delay", and I've not been able to totally eliminate it without additional workarounds (pre-delay or moving the MIDI/Audio trigger ahead). But every solution in Reason has SOME delay that (without workarounds) prevents gaining any headroom out of the operation.

[ShadowMatriks]

Good to know. I haven't tested all the options back to back!

I have tried Screams follower, M-Class and the Red Rock free multi comp thing, which has a minimum attack of 0.01 so in theory faster than 0.1 M-Class and SSL Bus Comp. Although I had a problem with it not resetting to zero. But this could just be down to how I had the combinator setup, as you need to do it unipolar but only allow 0dbfs to - numbers in the editor, otherwise it stalls on a value.

If you're saying Sweeper is the fastest Reason option I'll use that from now on thanks

[ShadowMatriks]

I thought Reason as a whole runs 64 frames late to allow for all this CV action. Is that the "SOME delay" you are referring to?

Any idea what the Fast attack time is of the channel comp?

[aquil]

AFAIR Reason adds a delay in some situation's to prevent feedback loops with 0ms delay, which would just generates a endless peak sound. The delay time is related to your buffer size.

I checked that again. If you use the KHS ring mod as an insert FX in a mixer track, 7 samples of delay are immediately added. So, it probably does come from the Rack Extension. Apparently, if you use audio splitters/mergers and manage to activate the LED on the back through complex routing to indicate non-standard routing, you can get Reason to no longer detect the delay. This, of course, means that nothing can be compensated via PDC.