Sample resolution and size question

[Steve Y]

Newbie to forum here. I recently had my new semi-customized Johannes Vivaldi 370 delivered and voiced, and it sounds wonderful. I asked the dealer about the sample size and resolution and all he could tell me was it was 24 bit depth but did not know the sample rate or length. In searching this forum as well as the net I could find sample size info for Hauptwerk (24 bit, 96 KHz, and long enough to capture all the reverb "tail"), but no information regarding samples for Allen, Rogers, Johannes or any others. I would not think such information would be proprietary but rather would be a good advertising claim as with the Hauptwerk system. However it seems they are keeping it a trade secret. Does anyone know where I can find comparative data from the major organ manufacturers?

[rjsilva]

Sampling resolution up to a certain point matters very little. The 24 bit 96 kHz advertised is a thing to promote VPO’s as a reason to believe they’re high quality. I think the most noticeable improvement is in higher frequencies such as with reverb or cymbals. But it’s not like a top notch 16 bit 44 kHz recording can’t sound excellent. And digitally generated reverb as found on organs nowadays would be at a high resolution anyway.

And then when you’re talking about an installation at a venue, whether a church or a large room, 16/44 vs 24/96 is going to mean squat (or less than squat!).

When I do recordings I use 96 kHz because of the natural room acoustics (I never do close up), but I’m certain that less than 1% of people could tell the difference through quality headphones. And when played through speakers I’d say less than 0.001% could tell. (I’m not meaning to be scientific with my figures...)

[ahlborn]

In the early years of 2000, companies publicized bit deph and sampling frequency with great emphasis.

In recent years, many companies have further upgraded their technology, so I assume that almost all of them use fairly high resolution/sampling rates that are no longer advertised.

However, I think these two factors are not the only elements that guarantee sound quality.

[don60]

I believe the OP was asking about the length of samples in seconds. This length from I can tell varies from rank to rank even within a given sample set--the loops are longer for pipe sounds that would show an unnatural repetition on a short sample.

Nowadays, the standard seems to be "several seconds." Back in the distant past, the ultimate in short samples was the original Allen MOS1 system that stored only a half cycle of a generic waveform and reproduced it at the required frequency using a clever calculation done in real time.

Traditional organ manufacturers probably do not want to divulge sample length since having "the longest" is of dubious technical merit in the first place and can easily be superseded by a competitor who wants to spend another $20 on memory.

[ahlborn]

There could be another interesting question.
Will companies that have increased bit/frequency have re-recorded all their sample libraries with higher sampling rates? Or did they use the same old samples?
I think it's hard to answer this question.
However, even oversampling old samples could (theoretically) have advantages. Even if the additional information can not be invented, oversampling could bring quantization noise very far from the audible range.

[Doddy.van.Straaten]

24-bit audio simply has a higher range of possible amplitude or level values than 16-bit, independent of frequency. This translates to a range of audible sounds of 144dB from the quietest sound to the loudest sound for 24-bit (which equates to roughly the maximum range of human hearing), 96dB range for 16-bit (CD quality), and roughly 50dB for a vinyl record.

So with a 24-bit recording you could capture all the detail of a mouse sneezing at 1m followed by a jet taking off 50m away, while a 16-bit recording and a noise floor set at 0dB SPL (sound pressure level at the location of the microphone), would capture the mouse sneeze, but the jet would be all distorted and clipped (falsely being reproduced at only 96dB). Setting the noise floor high enough to accurately record the jet, say 140dB, means the quietest sound you could record would be 44dB SPL or approximately the ambient level of a quiet home.

In terms of playing back audio recorded in the above example, pretend you had an amazing home stereo that could accurately playback the full 144dB range of 24-bit audio, in a perfectly sealed anechoic chamber. You calibrate the speaker output so that the levels you hear match the levels of the point of recording. You listen to the 24-bit recording, hear the mouse sneeze, then hear the jet take-off.

Assuming you can still hear anything, you play back the first 16-bit recording - the mouse sneezes, then the jet takes off, but this time its only 96dB instead of 140dB, and sounds distorted due to clipping. You turn on the second 16-bit recording, and you hear the soft sound of noise similar to the level of your living room, no mouse sneeze audible, then the jet takes of at 140dB, deafening you again. With a vinyl record, the resulting playback is - with a noise floor of 0dB, the mouse is audible and at the right level, but the jet is only 50dB, and with a noise floor set to capture the jet at 140dB, the resulting background noise is at a whopping 90dB.

As far as listening to the 16-bit vs 24-bit, for music not piped directly into your brain via USB, you will likely never notice any difference between 16-bit and any higher bit depth. It's still best to record and mix in the highest bit-rate and sample rates possible, but the final output can be in 16-bit for any practical listening setup.

[Melos Antropon]

First off, Steve Y, congratulations on your new organ! I also own an Opus 370, and I love it.

Johannus does use different lengths of the samples according to price level, but as indicated by other answers above, that is more an advertising ploy than anything else. They let the sample sound for the interval they want, then, using an oscilloscope, they put the "tail" on where there is an exact recurrence of that pattern, so there is no obvious "gurgle" or "click" to the loop. Johannus generates reverb completely separate from the sound and blends it in later - right at the console (that's what your "Cathedral" knob does). The Johannus reverb system is head-and-shoulders above the Lexicon reverb, to my ears (Lexicon reverb has that annoying "BOING, Boing, boing" on decay). But do read the posts above my carefully - they are filled with good advice.

Tony

PS: Where are you located, and what "customization" did you have done? I had them remove the 8 Salicional from mine and replace it with an Erzhaler Celeste II!

[jbird604]

Tony hints at what may help answer your curiosity, and that is the distinct difference in operating principle between Hauptwerk and other VPO's and a typical commercially-available packaged organ, whether it be Johannus, Allen, Rodgers, or whatever.

While all these are "digital" and all use "samples" to create the organ sound you hear, Hauptwerk and other VPO's depend on the recorded samples being VERY long, in some case perhaps several seconds. With this system, there is no need for much if any "processing" of the samples as they are played back. And "wet" sample sets even include genuine reverb tails or ambiance, so they sound precisely as they would sound in the original environment, often a major church or cathedral.

By contrast, all the commercial builders use their sample sets in a somewhat different manner. The actual length of the sample is not nearly as important to them, as they have sophisticated systems in their tone generation setup to provide much of the "liveness" that we expect to hear on organ tones. For example, Allen has long used a method called "Random Motion" to produce an artificial but highly realistic "jitter" in the steady-state tones. They started doing this with their original MOS digital organs, and it was then and continues to be quite effective in simulating the intended effect. When using this truly "random" method, they get what sounds like a genuinely endless "loop" from even the shortest sample, because the sample waveform is varied endlessly and with endless variety by the jitter circuit.

No doubt other builders so something similar to their tones, though some seem to be saying in their advertising that they use rather long samples and some even seem to claim that they have an actual long sample for every note of every rank. But I suspect most of today's digital organs are more like Allen's Renaissance, with samples that are perhaps a half second at the longest, and often just a few samples are "stretched" to cover the entire keyboard range. I could be wrong about that, and maybe someone else will chime in and say that they are certain that such and such an organ does indeed have 61 complete samples for every rank.

And few if any commercial digital organs use "wet" samples. Their reverb or ambiance is always generated by a separate system or perhaps a software routine and does not depend on having any ambiance present in the samples themselves.

Anyway, the difference between commercially-built organs and the world of VPO's is quite vast, and one area where that is obvious is in the sample length.

[don60]

[S]ome even seem to claim that they have an actual long sample for every note of every rank. But I suspect most of today's digital organs are more like Allen's Renaissance, with samples that are perhaps a half second at the longest, and often just a few samples are "stretched" to cover the entire keyboard range. I could be wrong about that, and maybe someone else will chime in and say that they are certain that such and such an organ does indeed have 61 complete samples for every rank.

Johannus uses a distinct sample for each note of each stop on their higher-end processors and has done so for many years. This feature is very evident in playing individual notes of individual ranks using the factory default finishing--flaws and variations in the original recordings stand out from one note to the next. One could say that the ultimate limitation of the Johannus technology is the pipe organs themselves that form the basis of their sound.

Viscount gets around this problem entirely with Physis, generating the sounds in real time.

[Steve Y]

Thank you everyone for the thoughtful replies and insight, what a welcoming group. A little background will make the purpose of my initial question a bit clearer. I own a recording studio and generally record folks into a ProTool HD system at 24 bit / 96 KHz. I do a lot of critical listening in the studio to the live and recorded feeds though a large professional desk and monitors. I realize that bit depth, sampling frequency and sample length are but three of many, many variables impacting the final sound. As was pointed out in several replies, looping, stretching and bit dithering, when used, contribute a great deal to the final sound. Another variable is the playback system of amps, reverb and speakers, which vary greatly, and of course the source organs may make the biggest difference. I live in SE Pennsylvania, about 15 miles from the Allen factory. It was always my intention to buy an Allen organ for my home, the build quality is second to none, and I would be supporting a local business. However when I auditioned organs from the various manufacturers I came away with a clear perception the Johannes instruments sounded the most realistic. This left me wondering about the technology differences between the companies. In the end I went with Johannes based on the sound, not the console construction.

[John Kinkennon]

A couple of clarifications. A 24-bit depth is required, but not because we need a signal to noise way in excess of 100dB. With large registrations, reverberation, and rapid playing it is normal to have well over 1000 audio samples combining at one time. Various sample set producers will suggest polyphonies of 4000 or 6000 to do the job! When audio samples are combined the noise power adds 3db with each doubling of samples. So if we do the math a super audio interface with a S/N of 115dB or so rapidly builds up audible noise with 16-bit samples.

Longer samples of several seconds result in less audible looping artifacts and a greater variety of phase relationships to prevent a stagnant sound. That said, putting many extra hours into picking looping parameters can result in shorter samples with the same overall quality. I have the Goerlitz sample set for Hauptwerk with 86 ranks if I recall. It utilizes 6 channels as well so the memory requirement exceeds 70GB. Manufacturers aren't likely to want to make comparisons. The downside is a couple of minutes to start the organ even with a Samsung 970 M.2 SSD. You don't get something for nothing.

I'm looking forward to the Organteq product which will be based on physical modeling. Their Pianoteq product is awesome so I think Hauptwerk will face an interesting competitor. I plan to use both.

[Steve Y]

"PS: Where are you located, and what "customization" did you have done? I had them remove the 8 Salicional from mine and replace it with an Erzhaler Celeste II!"

Tony, I did not change any of the voices, but had Johannes make the console from red oak and stain it Washington cherry to match our music room book cases. I upgraded the keyboards to their wooden core "tracker touch" versions. I added several extra toe pistons, mostly for indexing through memories and replicating a few of the thumb piston functions. In addition I got the lighted, adjustable music desk and the SP12 pipe facade/speaker system which sits on top, but won't arrive for another few months.

[ahlborn]

I live in SE Pennsylvania, about 15 miles from the Allen factory. It was always my intention to buy an Allen organ for my home, the build quality is second to none, and I would be supporting a local business. However when I auditioned organs from the various manufacturers I came away with a clear perception the Johannes instruments sounded the most realistic. This left me wondering about the technology differences between the companies. In the end I went with Johannes based on the sound, not the console construction.

On this point, and on the overall quality of Allen, I had several diatribes on this forum.

In the past, many manufacturers such as Ahlborn, Viscount, Johannus etc. they built low/medium quality consoles.
Today, this is no longer like that. The biggest manufacturers use quality materials and assembly.
Your Vivaldi 370 is built just as well as an equivalent Allen.
Johannus's high-line organs (Monarke, Ecclesia) are built just as well (and robustly) as Allen.

I also think that the T9000 Johannus technology is immensely superior to Allen.

[voet]

...when I auditioned organs from the various manufacturers I came away with a clear perception the Johannes instruments sounded the most realistic. This left me wondering about the technology differences between the companies. In the end I went with Johannes based on the sound, not the console construction.

While I am not a techy, I have really enjoyed reading this thread. The combined wisdom on this forum is truly impressive. Steve Y, I would be interested in knowing which organs and models you auditioned, if you do not mind sharing this information.

[Steve Y]

Bill, I tried a three manual Allen Historique (IIIa ?) and a Rogers Artist series three manual organ, think it was their 589, it had internal speakers, but that was well over a year ago, so I'm not sure. I felt the Rogers sounded better than the Allen, but both left me wanting. It is tough to describe the qualities I felt were missing, both sounded OK but did not **** ** away. I also listened to a two manual Johannes Opus, but the Vivaldi was a big jump, possibly due to its more extensive amp/speaker system.