What is the deal with good and bad reverb?

The Lounge General Discussion
1

Since I’ve been into synthesis (still relatively new) I’ve heard repeatedly reviewers talk about how such and such device has great/passable/shit/not studio-quality effects, particularly focusing on reverb.

My limited knowledge of reverb is that it’s crafted through an algorithm of some sort. So is this really hard to program, or something? Why would a device have a poor reverb - especially if it’s digital?

This is particularly odd when devices differ within the same company. Roland device A might have a great reverb, while Roland device B might have a shit reverb. Haven’t Roland just got “the reverb algorithm” down, perfected years ago, and usable on every device? Or am I missing something.

I know there’s expensive reverb boxes/pedals, which makes me even more curious.

(Context: I’ve recently been rinsing this lovely track, which has reverbs in it the like of which I’ve never managed to create myself!)

9 Likes
2

possible answer: there are lots of creative decisions to make when coding a reverb and people have different tastes as to what constitutes a “good” reverb anyway.

If there was an objective “good” reverb, we’d probably be getting there by now.
Or everyone would just be using convolutions.
But realism isn’t necessarily what people want from a reverb.

10 Likes
3

hello, that would be the case if you think every devices has the same processor and is coded in the same langage , and also do you think a roland aira compact has the same processing power has a fantom ?

1 Like
4

My guess would be nicer sounding reverb uses more CPU so on less powerful boxes or those with more features, they may need to trade quality for cpu cycles

6 Likes
5

Well, your quesiton presupposes that I’m aware that reverb is CPU-intensive.

4 Likes
6

This - I often prefer “bad” or at least characterful reverbs. Tin cans, dumpsters, yes please. You can roll these in Max or PD or whatever with a few delay lines and clever feedback/mixing.

Modulation helps take some edge off too (or at least makes it weirder), and you can’t get this out of convolution.

7

A lot of it is pretty subjective.

Reverbs tend to fall victim to the whims of fashion more than a lot of other effects, but ultimately, they’re a bit like snares, in that they’re harder to get right than you might think.

I’ve built a few reverbs in Max and they are pretty CPU intensive once you start adding all the bells and whistles, but I’ve made shit reverbs that use a lot of CPU and decent ones that are much more economical.

My favorite reverbs tend to be from over 30 years ago, because I like aliasing and the quirks of lower sample rates/old converters.

It’s all about figuring out what you like in a reverb and finding ones that fit your requirements, because there’s so many out there, there is really about 10,000 out there for each of us.

17 Likes
8

Digitakt feature request: special edition with the Room Reverb replaced with a Dumpster reverb model.

should sell three or four of those

:wink:

7 Likes
9

The basic algorithm of a reverb effect is not a rocket science and not a concern todays CPU-wise (btw in Bitwig there is a nice Grid preset of a vintage verb made by using all-pass filters in a loop, I believe it’s the main technique), but algorithmic reverb is a rough approximation of a real thing, a pure mathematical model. There are hundreds ways of implementing or tuning it. Some simple math will just sound hollow unnatural (dunno, a Freeverb - classic implementation) and there are advanced ways to make it more real or nice…
The Impulse-based reverbs are totally different thing, they are the most natural even in basic implementation imo, but even today they are too CPU-intensive to be easily put into a pedal

2 Likes
10

Advise: give the most information that you can so we can help.
Because reverb is processing, and not necessarily comes first in the chain… we do need more context to help. What is your setup, what are the gears, do you have a mixer, is it live performance, is it studio production, do you have a daw, do you ask information on hardware reverb, software, from your gear, from your daw etc…

General approach (and not repeating what other said on subjectivity, tastes, performance-cpu etc)

There’s few reverb categories, I suggest you to read the history of reverbs… (it all started with theather
/auditorium where reverb was a tailored and treated space design to enhance what happening inside… it’s usually what’s hall reverb about…) so you will understand those families. Then I suggest you to find some advise on which reverb families work with what in terms of material… (spring on guitars, organs… plate on vocals, room on drums etc of course a bit of shortcuts here to go faster.) what convolution reverbs are as well…

Then understand that there’s reverb & mixing so understand dry/wet (why we record dry as f*** and process after), reverbs on instruments, reverbs on drums, reverbs on vocals, reverbs on SFX (which could be more experimental reverbs)

There’s also tonal manipulation with reverbs (a bit like EQing so in a very subtile way)… I think the developer of Valhalla write some stuff on reverbs as very passionate…

But it’s up to you to try all types of reverb, on a lot of materials to choose what best on what… also as every space processors you recreate on raw/dry materials which the ears didn’t like much (vocal without reverb mostly we don’t like it or at least it usually difficult to preview a voice dry in a mix) you have to make a connection with your room, headphones to not overdo the amount of reverbs usually in headphone we tend to put too much decay….

Reverb is a big topic more than we can think when we start.
Some mixing engineer use a lot of reverbs, some use techniques to position in 3 dimensional front - middle - back… it tends to go even more complex with mixing for Dolby atmos

Understand reverbs in depth can really help to reduce the amount of reverbs we have as well - and of course taste will play a role that’s for sure typically I use reverbs mostly as sends-returns (aux) and tend to not use internal reverbs because I like to have more control over it (EQ pre / EQ post / sidechain, distortion whatever…)

2 Likes
11

The guy from Valhalla DSP has a blog, worth reading if you’re interested. He covers the history of reverb and different approaches to it as well as different dsp eras.
I guess it comes down to these points: A “good” reverb is subjective and usually the taste of people involved in coding them is reflected in how it sounds and also the fine-tuning is often done by ear. So someone might actually tweak all the delays and allpass filters, how they mix etc. by ear. I’ve done that once and it’s incredible how much time you can spend doing that. It basically never ends.
Or if it’s not algorithmic, there are also many factors that go into the experience. Even the gui might drive one into a certain direction.
But also hardware/platform specs might play a role and taste is often a result of what music people like (=what others have done with reverb). Lots of variables that each play their role.
Look at 80s rack reverbs and how people are starting to appreciate the once poorly specced sound.

18 Likes
12

That’s what I have in mind when mention Valhalla thanks to confirm that

1 Like
13

Or we’ve gotten to where we have so much processing power it is “good enough” that people stopped caring about interesting musical kludges to approximate, and it’s so easy to devote resources that people don’t bother to look as deep into research papers for algorithms or recreating classic designs. Valhalla and a few other manufacturers the exception to the rule.

Edit: Beaten to the point while getting coffee.

@mitya33 as mentioned, check out Sean’s blog for some tidbits of history that will get into older gear that he’s reverse engineered the algorithms from or the basic process- Getting Started With Reverb Design, Part 4: The Best Books - Valhalla DSP

Then there’s an other vector of commodity dsp platforms based on older Ensoniq designs and their ease/limitations.

Some people use the basic in box algos (there’s a reason why a lot of boutique pedals sound the same!) some people roll their own.

5 Likes
14

Syntorial can help with your learning in synthesis and reverbs … as you train your ears … I remember to failed the panning & reverb exercices with headphones (all parts green at 100%) while on monitors I was mostly right… (but I think it’s pretty rudimentary meaning I don’t think you will go learn reverb types with syntorial it is mostly synthesis… and not specific to reverbs)
Experiencing those kind of things really help to progress though

Some ressources :
obvious Valhalla blog (link fixed)

Izotope post on Reverb’s history

LiquidSonics post on reverb’s history

Eventide on reverb types to associate with materials types

Another one on associate reverb types to audio materials

Some Great Courses :
Puremix on how to listen Reverbs

MixMasterWyatt : Mixing with Valhalla Reverb

Very basic but a good start point :
(and not really precise so it’s very to get a draft idea)

Capture d’écran 2025-04-14 à 18.16.03
Capture d’écran 2025-04-14 à 18.16.03808×930 313 KB

App to train your ears :
Syntorial
Train your ears on EQ

6 Likes
15

Echoing a lot of other people here, I think reverb isn’t a solved problem - processing power is still an issue. I think that goes double for things like synths and grooveboxes with relatively limited CPU power, especially when you might have to run multiple FX.

There’s no reverb on the market today that does what I want, I have used some of the best out there (eventide, arturia, valhalla, lexicon, liquidsonics, denise, the list goes on). Some of them bring my music PC to its knees and I still have to set up an FX bus with pre and post processing, and sometimes multiple busses with different reverbs that are routed in and out of each other, to get close to the sound I hear in my head. I’ve tried to program it myself and I just break whatever higher level interface I’m using (since I can’t actually code).

And all of that is just considering stereo reverb, but add in stuff like surround sound or atmos and it gets even more complicated (and CPU intensive). I think there is a real need for solid reverb that can adjust dynamically in real time for things like AR/VR and raytraced audio, and that will need more processing power and possibly even specialized processors to run reasonably. That’s more on a game and applications side of things than music, but that would probably have effects on music applications too.

1 Like
16

In this case it’s probably mostly available processing power, but could also be certain demands the hardware makes that warrants porting (=adapting the algorithm) it. So the same algorithm can sound different on different devices. You can try that in your DAW by changing sample rate. Now think it’s not only sample rate, but also bit depth, certain restrictions on what math can be done on a particular processor etc. etc.
In a hardware setting, there could be many different factors that end up changing how the reverb sounds. Even the headroom at the main out mixing stage could influence it. It might even sound nicer with some saturation and different elements fighting in the mix or maybe it ends up sounding less detailed
There’s also the corporate side of it; an entry level groovebox is expected to have a less detailed sounding reverb, so it might end up getting exactly that.

There’s also the psychoacoustic side…if you let people listen to a simple boring saw wave, one coming from a 150 euro groovebox and the other one from a 5000 euro synthesizer, most will prefer the expensive synth. Even if it was the same saw wave coming from a soft synth.

One could argue that impulse responses can already recreate any reverb perfectly, from real spaces to expensive reverb units, yet people still tweak their algorithmic reverbs and enjoy their 80s Quadraverb or whatever cheap mass produced reverb unit it is they love.

1 Like
17

I can’t read this title and not do Jerry Seinfeld voice!

15 Likes
19

Also automations are important key concept with reverb (and effects in general)
it can really change effects from passable as static to useful in the dynamics/time/vibes I would say
usually it’s a neglected approach.

4 Likes
20

You have the re, and you have the verb.

7 Likes
21

Some reverb is good, some reverb is bad.

But…it’s all subjective.

3 Likes