Eventide SP2016 Reverb

[jens]

The SP2016 by default is louder than the Stereo Room version and doesn't have the same level of adjustability in the EQ. Adjusting the factory presets to match loudness and EQ delivers a sound that is a little bit different but no more than that. And by "little bit" I mean "I'd fail a blind test on it".

The EQ is the same in both any either way I zeroed both filters in both plugin - and I clearly wrote "after compensating for the gain difference", so I really don't know why you feel a need to bring this up again.

Again: just look at their stereo-distribution in a phase-scope - and besides the audible difference between them both is by no means subtle. I heard it immediately and clearly , otherwise I would not have brought this up here in the first place.

[jens]

Not that the afore-mentioned Dan Gilllespie had anything to do with the development of the 2016 emulation (then or now - so I was wrong about this), thus he might not know every detail about it, but here is another quote from him:

I think it's most accurate to say that the reverb algorithm is the reverb algorithm and is the same between the 1980 SP2016, the 2004 Reverb2016, and the new plugin. The 2004 hardware ran on Motorola 56k chips and was written in assembly. The new plugin runs on your x86 chip and is written in c++. However, work was done to the plugin to mimic the specifics of the processor from the 1980 SP2016. I know it has to do with sample rates and the bit depth of summing junctions, but I don't know the full particulars. Additionally, work was done to reproduce the convertors.

Fwiw, this is usually the sort of thing that makes the difference between a good 2004 model and a good 2018 model.

[perpetual3]

Well, I think the SP2016 sounds great. I love the position parameter and the ability to automate it without artifacts. I accidentally also downloaded the stereo room 2016 plug and the SP2016 sounds much better to me.

[dermage]

Edit: Nevermind

All has been said.

[ghettosynth]

Not that the afore-mentioned Dan Gilllespie had anything to do with the development of the 2016 emulation (then or now - so I was wrong about this), thus he might not know every detail about it, but here is another quote from him:

I think it's most accurate to say that the reverb algorithm is the reverb algorithm and is the same between the 1980 SP2016, the 2004 Reverb2016, and the new plugin. The 2004 hardware ran on Motorola 56k chips and was written in assembly. The new plugin runs on your x86 chip and is written in c++. However, work was done to the plugin to mimic the specifics of the processor from the 1980 SP2016. I know it has to do with sample rates and the bit depth of summing junctions, but I don't know the full particulars. Additionally, work was done to reproduce the convertors.

Fwiw, this is usually the sort of thing that makes the difference between a good 2004 model and a good 2018 model.

So all of that backs up what I'm saying. As I say, as several have backed up, and as the facts that you post above show, they are both models of the same algorithm with advancements made only in the accuracy of modeling specific, and limited, aspects. I understand the changes that they've made, and frankly, some of the language above is meant to sound impressive to people who don't. They aren't actually modeling the CPU, only some details of timing with respect to the CPU. Such changes will yield subtle differences. Telling others that they need to listen more carefully is arrogant at best, but misses the point; if you have to listen carefully to hear the differences, then they are, by definition, not that significant.

[ghettosynth]

Well, I think the SP2016 sounds great. I love the position parameter and the ability to automate it without artifacts. I accidentally also downloaded the stereo room 2016 plug and the SP2016 sounds much better to me.

The position parameter is useful, it's also on the 2016 room.

[sqigls]

Not that the afore-mentioned Dan Gilllespie had anything to do with the development of the 2016 emulation (then or now - so I was wrong about this), thus he might not know every detail about it, but here is another quote from him:

I think it's most accurate to say that the reverb algorithm is the reverb algorithm and is the same between the 1980 SP2016, the 2004 Reverb2016, and the new plugin. The 2004 hardware ran on Motorola 56k chips and was written in assembly. The new plugin runs on your x86 chip and is written in c++. However, work was done to the plugin to mimic the specifics of the processor from the 1980 SP2016. I know it has to do with sample rates and the bit depth of summing junctions, but I don't know the full particulars. Additionally, work was done to reproduce the convertors.

Fwiw, this is usually the sort of thing that makes the difference between a good 2004 model and a good 2018 model.

So all of that backs up what I'm saying. As I say, as several have backed up, and as the facts that you post above show, they are both models of the same algorithm with advancements made only in the accuracy of modeling specific, and limited, aspects. I understand the changes that they've made, and frankly, some of the language above is meant to sound impressive to people who don't. They aren't actually modeling the CPU, only some details of timing with respect to the CPU. Such changes will yield subtle differences. Telling others that they need to listen more carefully is arrogant at best, but misses the point; if you have to listen carefully to hear the differences, then they are, by definition, not that significant.

[ghettosynth]

Did you make enough for everyone?

[jens]

https://www.dropbox.com/s/vvfgnubwmmfxd ... R.wav?dl=0

Same settings on both plugins, output of SP 2016 brought as close to 2016 SR as possible, normalized afterwards... just guitar and the reverb, no other processing... (you have to right-click)

Don't tell me you can't hear the dramatic difference between both... ?

[pmcculloch]

Eventide guy here. As someone who has spent a long time working with both versions of the plugin, as well as the Princeton Digital Reverb 2016 hardware box--and is fond of all of them--there definitely are differences between the products. How subtle or important they are is a question of perception and/or musical taste.

As a bit of background, the three main reverb algorithms from the SP2016 are known for their smooth tails and do not use any modulation to break up resonances; this means that their frequency response is static and you will not get modulation artifacts in your reverb tails. When he created the algorithms, Tony Agnello spent a long time carefully tuning them by hand. The most noticeable differences in the various iterations of the algorithms are how each handles the difference between the original box's sample rate and the plugin's sample rate.

2016 Stereo Room changes the length of its internal delays to accommodate different sample rates. This is an approach that is very efficient and especially made sense in earlier plugins where CPU usage was more limited, but it doesn't sound quite as close to the box as our new approach. It also means that 2016 Stereo Room will sound slightly different at different sample rates, not only because the room modes will have different distributions but also because the internal filters will not be identical. In what is essentially a giant feedback structure, small differences in details can produce big changes.

SP2016 Reverb processes the wet path at the original sample rate of the SP2016 (~39 kHz), and that makes it significantly closer to the original, though it also means that the plugin is more expensive because it has to do resampling on the inputs and outputs of the algorithm. SP2016 Reverb will sound the same at different incoming sample rates, because the internal sample rate is constant. We've also A/Bed the algorithm against one running at 40 kHz and there is a difference in the smoothness. This likely wouldn't be as significant in an algorithm which used delay line modulation, but these algorithms don't.

[Harry_HH]

Great work!

[Gamma-UT]

https://www.dropbox.com/s/vvfgnubwmmfxd ... R.wav?dl=0

Same settings on both plugins, output of SP 2016 brought as close to 2016 SR as possible, normalized afterwards... just guitar and the reverb, no other processing... (you have to right-click)

Don't tell me you can't hear the dramatic difference between both... ?

Well, those are different. However, there are a couple of issues here. The tail in one is noticeably more prominent, though it's not a pure gain issue. Most (but not all) harmonics get enhanced so the tail works out a bit louder and seems to bloom more. This maybe is related to the sample-rate/feedback interactions pmcculloch describes – I think I was only doing the comparisons at 44.1kHz, which I expect would lead to a relatively small difference versus the hardware's 39kHz.

Also, If one were matching sound rather than parameters, then the difference between the two reduces a lot (for the same algorithm). I was mainly comparing full wet vs full wet mostly on the Large Room as it makes the matching easier to pull off. Obviously, if you want the plate and other additional algos, then SP2016 seems the better choice now (especially at the intro/upgrade price I got it for at launch).

[Hermetech Mastering]

Eventide guy here. As someone who has spent a long time working with both versions of the plugin, as well as the Princeton Digital Reverb 2016 hardware box--and is fond of all of them--there definitely are differences between the products. How subtle or important they are is a question of perception and/or musical taste.

As a bit of background, the three main reverb algorithms from the SP2016 are known for their smooth tails and do not use any modulation to break up resonances; this means that their frequency response is static and you will not get modulation artifacts in your reverb tails. When he created the algorithms, Tony Agnello spent a long time carefully tuning them by hand. The most noticeable differences in the various iterations of the algorithms are how each handles the difference between the original box's sample rate and the plugin's sample rate.

2016 Stereo Room changes the length of its internal delays to accommodate different sample rates. This is an approach that is very efficient and especially made sense in earlier plugins where CPU usage was more limited, but it doesn't sound quite as close to the box as our new approach. It also means that 2016 Stereo Room will sound slightly different at different sample rates, not only because the room modes will have different distributions but also because the internal filters will not be identical. In what is essentially a giant feedback structure, small differences in details can produce big changes.

SP2016 Reverb processes the wet path at the original sample rate of the SP2016 (~39 kHz), and that makes it significantly closer to the original, though it also means that the plugin is more expensive because it has to do resampling on the inputs and outputs of the algorithm. SP2016 Reverb will sound the same at different incoming sample rates, because the internal sample rate is constant. We've also A/Bed the algorithm against one running at 40 kHz and there is a difference in the smoothness. This likely wouldn't be as significant in an algorithm which used delay line modulation, but these algorithms don't.

Thanks so much for the explanation, good to know what's really going on, and a case to be made for keeping both plugins installed! I've found some sources/sample rates where one sounds better than the other, and vice versa, so definitely good to try both and go with whichever sounds best.

[ghettosynth]

Yes, of course you can hear the difference in those samples. I can hear "A" difference, that difference isn't a "dramatic difference" when taken over many different kinds of settings. It's exaggerated because of the percussive source. I can hear "A" difference when I compare them as well, that doesn't change the fact that they sound "largely the same" to me. I think that we're just differing in opinion on what level of difference validates spending additional money. Both reverbs have a very similar character as every shred of evidence in this thread will tell you. Yes, if you have neither, get the new one, if you have the old one and you don't have a fetish for accurate reverb hardware emulation, then I still don't think that the new one is worth a lot of money and I wish that I would have waited until the next bundle update.

Now, if you crank up the sample rate, the difference IS exaggerated and the original sounds pretty bad. But at 44.1, it's not enough that I care TBH.

[jens]

I used these strokes on purpose of course so one can hear the tail better... in my music I tend play a lot of percussive bits here and there too though, so it's not as if this was a completely unrealistic example...