ValhallaPlate Updated to Version 1.5.0

[foosnark]

Demo'ed this and fell in love. Didn't like the room when I demoed it a long time back.

But seeing as Room seems to be the most popular choice, perhaps I overlooked something?

As a reverb layman, might I ask - which one should i get? I only have budget for one, so which one should that be? I want it on pads, vocals, acoustic piano, and maybe guitar.

Just to make the decision more difficult, have you tried Vintage? It's also great.

Room is quite good, but I personally like Vintage and Plate more. Difficult to say which of those two I'd choose if I could only have one. It'd take an epic reverb battle to decide that.

[Leonardus]

Hi Sean. There are a handful of products in the hobbyist/home project studio world that, when they were released, shocked the world. Their quality at their price was such that they were unquestionably purchase worthy and had everybody grinning ear to ear. The original Line 6 Pod, Rode NT1, RNC 1773 and your Vintage Verb are among those rarefied few.

Not too long ago you released an update to Vintage Verb that included new plate algorithms-(which play very well with my '80's synths . Would you please take a moment to describe the similarities and differences to these and your new release? Thank you Leonard

[valhallasound]

Hi Sean. There are a handful of products in the hobbyist/home project studio world that, when they were released, shocked the world. Their quality at their price was such that they were unquestionably purchase worthy and had everybody grinning ear to ear. The original Line 6 Pod, Rode NT1, RNC 1773 and your Vintage Verb are among those rarefied few.

Thank you for your kind words!

I gotta admit, my goal for ValhallaPlate was to come out with the reverb equivalent of the SM57/SM58. Cheap, easy to use, versatile, no need to think about it. Just stick it in front of the singer/amplifier, and go.

Not too long ago you released an update to Vintage Verb that included new plate algorithms-(which play very well with my '80's synths . Would you please take a moment to describe the similarities and differences to these and your new release? Thank you Leonard

The "plate" modes in VintageVerb were all based on "digital" plates. More precisely, they were based on different older Lexicon algorithms, with significant changes and modifications (as well as the indeterminacy of not knowing exactly what went on in those old boxes).

• Plate: closest to the Plate in my PCM70. A good sound, but kinda choppy at low decay values.
  
  Dirty Plate: based on some time spent with the Plate algorithm in the 224XL. This is a more EXCITING sound than the other VintageVerb plates. There is some definite metallic coloration, and a bit of left/right bounce. Really nice for drums.
  
  Smooth Plate: I originally called this "Rich Plate," as I fell in love with the Rich Plate algorithm in the 224XL. I wasn't able to capture this exact sound, but Smooth Plate turned into something different. Much bigger maximum "Size" than the other plate algorithms, so the modal density is higher. The reverb decay is shaped so that it has a much smoother decay than Plate or Dirty Plate, and the input is shaped to reduce transients. I've heard this algorithm described as sounding "expensive." I don't like the class implications of that description, but I know that Smooth Plate sounds good on most any source material.

The ValhallaPlate algorithms have pretty much nothing to do with the VintageVerb algorithms. Totally different topologies, totally different reverb school.

• The ValhallaPlate algorithms are designed to emulate steel plate reverbs, with instant echo density, while avoiding the metallic artifacts associated with the Lexicon algorithms at high diffusion values.
  
  There is a LOT more shaping of the reverb tail in ValhallaPlate, in order to have a smooth exponential decay, that has different decay rates in different frequency bands.
  
  A lot of work went into avoiding the high frequency "hash" that can be heard in a digital reverb, when high frequency transients are sent into a long decay.
  
  ValhallaPlate has dispersion, to warp the sound of the initial reverb attack at different frequencies, and to create a wider & deeper stereo image.

I feel that ValhallaPlate is a far more "natural" reverb than VintageVerb. Meanwhile, VintageVerb sounds "digital," which is a sound in and of itself. My reverbs are like my children - I love them all!

Sean Costello

[Stopani]

I shall be grabbing this at the weekend. Great to see more stuff coming from ValhallaDSP.

[vic_france]

My reverbs are like my children - I love them all!

Sean Costello

I heard that you sent VRoom off to boarding school, and it is still waiting for a food parcel from you

[valhallasound]

My reverbs are like my children - I love them all!

Sean Costello

I heard that you sent VRoom off to boarding school, and it is still waiting for a food parcel from you

ValhallaRoom is getting a GREAT education! Hazing builds CHARACTER!

[Liero]

I _REALLY_ like that there are fewer controls and knobs on this thing, might just be me though.

[Krakatau]

Smooth Plate turned into something different. Much bigger maximum "Size" than the other plate algorithms, so the modal density is higher.

Sorry, what does modal density means exactly ?
(obviously nothing in common with modal transposition, AFAIK...)

I'm surprised how well ValhallaPlate does on the "subtle diffusion" ambiences, since it wasn't designed for that.

The nice thing for us, costumers is that we might perhaps, have a chance to gain some more algorithms dedicated to this peculiar aspect in a future update of VP !

My reverbs are like my children - I love them all!

Then, be just a happy "père de famille nombreuse" !!!!!

[valhallasound]

Smooth Plate turned into something different. Much bigger maximum "Size" than the other plate algorithms, so the modal density is higher.

Sorry, what does modal density means exactly ?
(obviously nothing in common with modal transposition, AFAIK...)

Modal density = resonance density = eigenmode density. IT'S JUST THAT SIMPLE, PEOPLE!!!!!!!



OK, for those people that don't eat, breathe, and sleep reverb design, a bit of explanation:

• A room is often described as having resonant "modes."
  
  Each of these modes resembles a sine wave, with an exponential decay.
  
  These modes are "excited" by signals that are close to them in frequency.
  
  When you hear the reverb of a space, you are essentially hearing these modes ringing out.
  
  Imagine a room as being filled with lots of tiny little bells, each of which has a specific resonant frequency.
  
  When a musical note rings in the room, only those bells that are close to the frequencies in the musical note will ring out.
  
  The larger the room, the more resonances filling that room.
  
  A concert hall might have a few BILLION parallel resonances.

I'm talking about rooms, but the same idea of resonant modes can be found in any resonant object. Plates, springs, piano soundboards, guitar bodies, what have you. Strings can be expressed as modes as well.

For example, a piano could be simulated with a whole bunch of second-order resonators, each of which produces a decaying sine wave (or damped exponential, if you want to get fancy). The soundboard has a whole bunch of resonances, kinda randomly distributed. They have really short decays. Meanwhile, the strings all have really LONG decays. They will ring out for far longer than the soundboard resonances.

The piano is a good starting space for where we are going next. As a kid, I used to play around with our old upright piano as a reverb. Hold down the sustain pedal, and sing into it. Your voice will excite all the strings that are close to the pitch of your voice. It makes a pretty cool reverb.

Now, imagine doing the same thing, but you are only holding down certain black keys when you sing into the piano. In this case the piano won't reverb everything - just the pitches that your voice and the black keys have in common. It won't sound like a proper reverb, just some weird chord.

Next, sing into the piano, and hold down middle C and the B right below that. The only pitches that will ring out are B and C. Total dissonant semitone. Not a good reverb at all.

The resonant modes in a room are similar to the strings in the above model:

• You want a BUNCH of them, to cover a wide enough range
  
  You want the modes to be well distributed, so they don't ring out in a "chord"
  
  You want the modes to decay at roughly the same rates in a given frequency area, so you don't have one or more hanging out at the end of the decay
  
  You want to avoid weird beating sounds when two modes are just the wrong distance from each other

By having a high density of modes, you increase the chances that several of them will be "excited" by a given input signal. If only two modes are excited, they will beat against each other with an unpleasant pattern. Once more than 2 modes are excited, the beating pattern gets more complex. Ideally, you want kind of a random beating pattern, as this sounds "reverberant" to our ears. A higher modal density helps get this random beating pattern. So does modulation.

This is assuming that you WANT a high modal density. For something like a steel plate reverb, you may want a lower modal density, as real plates have a fixed modal density (I've read that 1.17 modes/Hz is the statistical average, but different plates will have higher or lower averages). By turning the SIZE parameter down in ValhallaPlate, you are reducing the modal density, so you can get more of the ringing effects found in a plate. The modes in a plate should all decay fairly evenly, but there will be some audible beating effects on certain input signals.

Sean Costello

[Hermetech Mastering]

Thanks for that explanation, Sean, have often wondered exactly what you were talking about when you mentioned modal density, and now we know. Should definitely put that up on your blog!

[egbert]

Very interesting explanation, Sean.

When I think of resonances I think of standing waves - eg in a room shaped like a rectangular prism you have a set of frequencies where the reflections back and forth between the parallel walls will reinforce (creating a standing wave) and other intermediate frequencies where they will interfere destructively and create nodes. At the low end of the spectrum these peaks and nodes are well separated.

I was surprised that a rectangular plate with two sets of parallel sides - which is a simpler case than three sets of parallel walls with different separations - would have a density of modes more that 1 per Hz across the audio band - does this reflect the density of modes in the higher frequencies while there are fewer in the bass? I realise that the speed of sound in a metal plate under tension is much faster than in the air in a room.

[Krakatau]

Modal density = resonance density = eigenmode density. IT'S JUST THAT SIMPLE, PEOPLE!!!!!!!



OK, for those people that don't eat, breathe, and sleep reverb design, a bit of explanation:

• A room is often described as having resonant "modes."
  
  Each of these modes resembles a sine wave, with an exponential decay.
  
  These modes are "excited" by signals that are close to them in frequency.
  
  When you hear the reverb of a space, you are essentially hearing these modes ringing out.
  
  Imagine a room as being filled with lots of tiny little bells, each of which has a specific resonant frequency.
  
  When a musical note rings in the room, only those bells that are close to the frequencies in the musical note will ring out.
  
  The larger the room, the more resonances filling that room.
  
  A concert hall might have a few BILLION parallel resonances.

I'm talking about rooms, but the same idea of resonant modes can be found in any resonant object. Plates, springs, piano soundboards, guitar bodies, what have you. Strings can be expressed as modes as well.

For example, a piano could be simulated with a whole bunch of second-order resonators, each of which produces a decaying sine wave (or damped exponential, if you want to get fancy). The soundboard has a whole bunch of resonances, kinda randomly distributed. They have really short decays. Meanwhile, the strings all have really LONG decays. They will ring out for far longer than the soundboard resonances.

The piano is a good starting space for where we are going next. As a kid, I used to play around with our old upright piano as a reverb. Hold down the sustain pedal, and sing into it. Your voice will excite all the strings that are close to the pitch of your voice. It makes a pretty cool reverb.

Now, imagine doing the same thing, but you are only holding down certain black keys when you sing into the piano. In this case the piano won't reverb everything - just the pitches that your voice and the black keys have in common. It won't sound like a proper reverb, just some weird chord.

Next, sing into the piano, and hold down middle C and the B right below that. The only pitches that will ring out are B and C. Total dissonant semitone. Not a good reverb at all.

The resonant modes in a room are similar to the strings in the above model:

• You want a BUNCH of them, to cover a wide enough range
  
  You want the modes to be well distributed, so they don't ring out in a "chord"
  
  You want the modes to decay at roughly the same rates in a given frequency area, so you don't have one or more hanging out at the end of the decay
  
  You want to avoid weird beating sounds when two modes are just the wrong distance from each other

By having a high density of modes, you increase the chances that several of them will be "excited" by a given input signal. If only two modes are excited, they will beat against each other with an unpleasant pattern. Once more than 2 modes are excited, the beating pattern gets more complex. Ideally, you want kind of a random beating pattern, as this sounds "reverberant" to our ears. A higher modal density helps get this random beating pattern. So does modulation.

This is assuming that you WANT a high modal density. For something like a steel plate reverb, you may want a lower modal density, as real plates have a fixed modal density (I've read that 1.17 modes/Hz is the statistical average, but different plates will have higher or lower averages). By turning the SIZE parameter down in ValhallaPlate, you are reducing the modal density, so you can get more of the ringing effects found in a plate. The modes in a plate should all decay fairly evenly, but there will be some audible beating effects on certain input signals.

Sean Costello

First of all another big Thank you for the concise but detailed and very instructive explanation,

Talking about modal density, i assume that the creation of selective resonances, bounded to a specific scale or another will be nothing but a bunch of delay lines of very short delay times
AFAIK, equal to (or less than) 50 millisecond in the way that a wave of 20 hz (or more) at the low-end of the audible bandwidth will enter in phase with it's own delayed signal, it feedback included

the unproperly called "Comb filters" that will more or less emulate the behavior of sympathetic strings

So in my mind, the question is, would it be another way using algorithmic process to create selective resonances density, bounded to a scale or another, something in between these comb filters and more traditional algorithmic reverbs

I have a personal concern on the peculiar case, because it falls right on my ballpark in the convolution counterpart : https://www.dropbox.com/s/qk6fbnkbbciak ... t.zip?dl=0
...but of course, with the inherent flaws of the IR technique (static responses and limited flexibility) although with a variety of timbres that APPARENTLY can't be rivaled by algorithmic methods, so if you feel concern, you might perhaps have an idea if it would be possible to go over these boundaries ?
_____

P.S. the download being rather big (about 1.5 Gig) this in and mp3 example of the use of these scales-bounded resonances : https://soundcloud.com/perbuatan1883/gmelmintutorial

[4damind]

Is the "width" parameter only some M/S thing or is this some more special "delay" or micropitch-shifting between L/R? (Anyway, it sounds very good also with 200%)

[valhallasound]

Is the "width" parameter only some M/S thing or is this some more special "delay" or micropitch-shifting between L/R? (Anyway, it sounds very good also with 200%)

M/S, but above 100% it only works on the low frequencies.

[Andrew Souter]

• A room is often described as having resonant "modes."
  
  Each of these modes resembles a sine wave, with an exponential decay.
  
  These modes are "excited" by signals that are close to them in frequency.
  
  When you hear the reverb of a space, you are essentially hearing these modes ringing out.
  
  Imagine a room as being filled with lots of tiny little bells, each of which has a specific resonant frequency.
  
  When a musical note rings in the room, only those bells that are close to the frequencies in the musical note will ring out.
  
  The larger the room, the more resonances filling that room.
  
  A concert hall might have a few BILLION parallel resonances.

..in other words, imagine it is like Kaleidoscope, but instead of up to 512 "spring resonators" you might have millions, billions etc... this is what reverb is like... if we talk in terms of "modes"...

mode = exp decaying sine wave = spring resonator

physical objects, strings, plates, rooms, etc tend to have harmonics (or more accurately "partials") for these "fundamental modes" also. So they are more efficiently modeled using Comb filters which give the fundamental plus all harmonics... (and much more complicated structures involving these kind of things...)

A single comb filter at 20 hz will give you 1,000 modes between 20 and 20,000hz.

500 parallel combs for example tuned to an exponential function tuning curve with a range of say one octave and a base freq of say 10 hz will give you something like ~30,0000.

make this 1hz instead of 10 and now you have ~300,000... it will sound like a semi-sparse reverb.

make this much higher like 100hz, and now it will sound like a string/bell/metal object... with very strong pitch. you will have ~3,000 "modes"... at 1000z you will have ~300.

consider a single comb at 1/2 Nyquist = 1/4 sample rate = 11025hz for 44.1k is the same as a single sine at that freq! since the next harmonic is "out of bounds" in the spectrum already...

really you will have infinitely more in all cases if we are talking about physical objects in the real world, but since digital audio has finite freq response we can only go as high as nyquist... and in the real world we can't hear ones above 20k or less anyway, so we don't really count them...

etc.

to be clear, more modes is not always better. it is better if you want to create a perfectly flat spectrum = white noise, which is usually part of the goal in "normal" reverb. Sometimes however you want comparitively few modes if you want a tonal resonator FX (like KS for example) where modes might be intentionally tuned to musical scales or anything else interesting...

don't mean to hijack.. just thought it'd be interesting to some people...