It is currently Fri Mar 29, 2024 5:21 am


When will we reach this level?

Existing and forthcoming Hauptwerk instruments, recommendations, ...
  • Author
  • Message
Offline

1961TC4ME

Member

  • Posts: 3144
  • Joined: Mon Feb 07, 2011 1:45 pm
  • Location: Lake Minnetonka, Minnesota

When will we reach this level?

PostFri Jul 15, 2022 3:10 pm

Hello all,

Been awhile since I've been on the forum here. I've had Hauptwerk since about 2010, currently have v5, and I plan to stay there with really no plans to update anytime soon, for the most part for what I write below.

Take a listen to the recordings below.

https://www.pipedreams.org/episode/2022 ... ensational

The instruments in this recording aside, just listen to the recordings. Use your headphones, your stereo system, however you like. Don't take any of this the wrong way, I applaud what's been accomplished and I've enjoyed Hauptwerk immensely over the years, but listen to all or just part of the recordings, and I just want to ask: Can anyone honestly say we've reached the level of clarity and sound we hear in these examples regardless of the HW version or instrument? Is it the way the recordings are being done? Is it the limitations we're up against using software attempting to reproduce the real thing? I don't know. Over the years I've tried about a million different audio layouts and routings schemes and have written about them here, some with great success, not to mention all the other things I've tinkered with to get the most realism I can get, and I'm just not hearing it. So, I've stopped and am calling it good and leaving it alone.

I know this is kind of an apples and oranges comparison, but I'm going on 10+ years now with Hauptwerk, I've updated 3 times from HW3, then to 4, and now to 5. There are now many, many instruments available, a number of them I've bought, but I've stopped there as well as until I hear what I hear in the recordings, I really see no reason to go further because to me it's just diminishing returns. I'd love to upgrade to 7, but I see no reason and will stay with 5 until I hear what I hear in these recordings.

Funny, but to my ears the most realistic sounding instruments I own and have heard to date are the Paramount instruments, add a hair of reverb and they're even better. AND they're all 16 bit recordings which is supposedly 'old school.' Go figure.

Marc
Offline

josq

Member

  • Posts: 912
  • Joined: Tue Jan 24, 2012 12:11 pm

Re: When will we reach this level?

PostFri Jul 15, 2022 5:32 pm

This question is hard to answer indeed, when comparing apples and oranges.

But for sample set producers it should be doable to check what level of realism they achieve. They just have to play a (simple) piece on the real organ using the same recording positions &conditions that were used for sampling. Then play back using some generic media player software and Hauptwerk respectively, using the same device (headphones/speakers).

If everything is done properly, a single sample of a single pipe played back in Hauptwerk should sound exactly the same as a recording of the same pipe.

Of course, we don't play single pipes/samples but chords & large registrations. This can be compared in a similar way. A lot will depend on the quality of the wind modelling & detuning in the sample set (and possibly some additional modelling like pipe coupling, swell box effects etc)
Offline

mnailor

Member

  • Posts: 1602
  • Joined: Sat Nov 30, 2013 5:57 pm
  • Location: Atlanta, GA

Re: When will we reach this level?

PostFri Jul 15, 2022 7:00 pm

This is very instructive listening. Thanks for the link!

My impression is that Hauptwerk 5 -7 *could* produce what I hear on these CDs, but the sheer sensation of spaciousness in attacks of the larger organs isn't there in any of the samplesets I have. There are other differences, but to me, that's what stands out.

A chord played on a big registration develops through time after the instant of attack because different pipes are at different distances from the mics or your ears. Even in a nearly dry room, if the organ is spread across multiple cases/chambers or a huge single case.

Samplesets often seemingly edit those varying distance-based silences off the front, so all the pipe speech starts at the same instant in the sampleset. But you'd only hear it that way from a very compact pipe organ or an electronic organ. This throws away a whole depth (= time) dimension of the spacial impression that's present in CDs and real life.

To be clear, I'm talking about the time it takes the pipe's attack to reach your mics/ears, minus the shortest time it takes any of the pipes' attacks to reach your mics/ears. That time delay is proportional to the distance to each pipe, and you hear those time delays as spacial position of the pipes, in addition to stereo position. If the varying delays are missing, it's not as believable.

HW could conceivably add random pipe delays, but the distance information is missing from the samples, so that would be just another "effect" like random detuning.

In the Goch sampleset, PG added a pipe delay option. I don't know if the option restores originally-recorded attack delays that were edited out, or introduces random or or otherwise made up delays. It does help realism.

I think Armley does a better job than most at having a big chord attack develop over time. Possibly Salisbury, too, but I haven't played it lately so that may be wishful memory.

(There's also preserving the mix of slower and faster speaking pipes, but I think most samplesets these days don't go too overboard editing those away. Also, the wind sag on the Midmer-Losh isn't usually what you get (or want) with large samplesets, but at least it's in the model.)

It should be possible to cut the exact same, shortest silence interval from every pipe's attack for the same mic position, so that the varying distance to pipes relationships stay as recorded, but the initial latency before the closest pipe speaks is eliminated so users don't gripe about a built-in delay. Make it an optional setting for people who want perfectly simultaneous pipe attacks?

Of course, dry/close samples where the mics are moved to in front of each chamber can't preserve distances. That's okay -- I don't think they serve the same purpose as ambient and rear samples do. The latter should be able to mix into a CD that's very similar to an original recording in the building at an intermediate listening position. But the pipe attack delays would have to be preserved to get close.

(Since Dudelange is the only one of these which has been sampled, and I don't have it, it's hard to make anything but very general statements...)

Yes, I've ranted about this issue before. Getting old and repeating myself.
Offline

larason2

Member

  • Posts: 752
  • Joined: Thu Feb 04, 2016 9:32 pm

Re: When will we reach this level?

PostSat Jul 16, 2022 7:50 am

Interesting discussion, and I’ve posted before about the gap between Hauptwerk and a real organ re: sampling pipes together vs. individually. On this and other issues, it seems clear from sample set producers that closing the remaining realism gap is either impossible, or not worth the time and effort needed to do it. In some matters, I personally think that too much realism is a bad thing, for instance if the quiet stops on a set are too quiet relative to the loud stops. On a real organ, this huge dynamic range usually isn’t a problem, but on a sample set, I’m usually annoyed at having to adjust the volume all the time. Speakers can’t reproduce a large dynamic range as well as real pipes. When I make CODM files, I usually tighten up the dynamic range a bit to avoid this. It also annoys me if a set includes pipe speech defects or room resonances that would be present on the real organ. So in my opinion a sample set is always in a way an instrument based on a real one. Modern sets have different approaches to sampling, channels, denoising, etc., but I still find myself playing old sets a lot, such as St. Georgenkirche or the Kappel Schnitger. Maybe they don’t sound as “realistic,” but they are still lovely instruments, and I find them inspiring. In my opinion, the sample set producer is also an artisan, and the sample set their work of art, and while a sample set may not perfectly replicate reality, it can still be a great instrument and sound gorgeous. I might experiment with “realistic” pipe delays, but it would probably annoy me after a while, as do any sampled ranks where the speech is too slow. My Allen used to have a “delay” setting for some stops, and I never used it, the delay just annoyed me.
Offline

mnailor

Member

  • Posts: 1602
  • Joined: Sat Nov 30, 2013 5:57 pm
  • Location: Atlanta, GA

Re: When will we reach this level?

PostSat Jul 16, 2022 12:48 pm

We all might have different, valid, expectations. Anything from playing live and hearing it as if at the console, to playing live but hearing as if out in an audience position or listening to an immediate CD, which is what I like.

The distant listening while playing at the console requires some willing suspension of disbelief: I hear console key and stop noises close up and instantly, but I hear the pipes from a distance out at the mic positions, but too soon. It would be more like in real life having headsets receiving live from the mics and the console were sound-insulated, but the distance delay would be bad. So the delay gets tossed to fix that, making it more unrealistic but more usable.

The initial attack time due to distance from the closest pipe to the mic is omitted deliberately, and I wouldn't want it any other way. I practiced on a small organ with a console 50 feet from the organ case long ago, and the delay was hard for me to tolerate. I want the timings on pipe attacks to preserve distances among the pipes, but without the time due to the minimum distance from pipes to the mics . All just a convenient fiction to make it playable but sound real from an audience position.
Offline

brooke.benfield

Member

  • Posts: 183
  • Joined: Fri Apr 10, 2009 10:38 am
  • Location: Oregon City

Re: When will we reach this level?

PostMon Jul 18, 2022 1:28 pm

The sanctuary I play in has multiple resonant nodes. I've done my best to voice around the issues but it is still a problem.

Flutes seem to suffer more in this environment than other stops. Sometimes while playing I've been so annoyed by this extra resonance, I've held on to the offending note and dropped the others so I can search for the source by canceling and restoring stops one at a time. When I find the guilty contributor, sometimes when I restore it the extra room resonance does not return or is at least different. Why? I think the reason is I managed to restore the stop out of phase with the other stops that were still playing.

I relate my experience here to support mnailor in what he is saying about time domain information being removed from the samplesets. I think removing this information is a significant omission and would like to encourage sampleset producers to explore techniques whereby it could be retained.

Getting a reliable trigger from a purely mechanical instrument would likely be quite a challenge. Perhaps MIDI out on an instrument so equipped could be used.
Brooke Benfield
Organist, Gethsemane Lutheran Church
Portland OR
Offline
User avatar

Jan Loosman

Member

  • Posts: 380
  • Joined: Sat Dec 20, 2008 4:33 pm
  • Location: The Hague, Netherlands

Re: When will we reach this level?

PostMon Jul 18, 2022 3:32 pm

I taste in some answers that some of the old skool sample sets (stereo or 4channel i guess) sound natural and often very musical.
Why is that?
When i purchased my first 6 channel set (Brasov) i was disappointed how it sounded. To my ears dull and uninspiring.
I mailed this to Jiri. Later 6 channel sets improved in this respect.
Reading the previous answers about different pipe delays to achieve a impression of space and to locate the position of the different pipes ,then i wonder if adding an extra spatial cue (perspective) or even adding 2 spatial cues (8 channel) to the front channels introducing other different timings, will do any good in this respect???. I guess not.

Jan
Offline

josq

Member

  • Posts: 912
  • Joined: Tue Jan 24, 2012 12:11 pm

Re: When will we reach this level?

PostMon Jul 18, 2022 4:05 pm

Jan Loosman wrote:I taste in some answers that some of the old skool sample sets (stereo or 4channel i guess) sound natural and often very musical.
Why is that?


It is often hard to look in the sample set makers kitchen. But things like noise filtering, release markers, pitch correction and wind modelling are very critical.

If a sample set sounds dull and uninspiring, maybe the noise filtering was overdone. If the acoustics sound wrong, maybe the release markers are not correct. If plenum registrations sound static and liveless, maybe the pitch is too perfect. If full organ registrations lack impulse and movement, maybe the wind model is not right.

My guess is that many sample set makers have attempted to automate and standardize their processes, occasionally at the cost of quality.
Offline

mnailor

Member

  • Posts: 1602
  • Joined: Sat Nov 30, 2013 5:57 pm
  • Location: Atlanta, GA

Re: When will we reach this level?

PostMon Jul 18, 2022 6:33 pm

I usually mix the front-diffuse, middle-distant, and rear mic positions in 6 - 8 channel samplesets, and only include the close-dry samples if they're needed for clarity. I don't think any of the recent 6 - 8 channel organs I have are musically worse than the older 4 channel surrounds, but I've stopped buying from producers whose samplesets have been disappointing, so it's a limited subset.

I do like being able to choose my mix with perspective sliders, e.g. the new Doesburg is a lot better for me than the old one without sliders, even though I'm still only using diffuse and rear samples and omitting dry.
Offline

1961TC4ME

Member

  • Posts: 3144
  • Joined: Mon Feb 07, 2011 1:45 pm
  • Location: Lake Minnetonka, Minnesota

Re: When will we reach this level?

PostTue Jul 19, 2022 10:25 am

Hello all,

I figured I'd let this one brew for a few days and see where it goes. Glad to see some good input on this subject. I started out wondering if perhaps Mr. Barrone of Pipedreams was just better at recording organs than the average guy :mrgreen: as you can listen to just about any of his shows and again you hear something there you don't hear in sample sets. He does use CD's in his shows along with recordings he's made in his travels as well and he's very skilled at recording organs.

Someone mentioned Armley Schulze and I will agree, to date I feel this is one of the best at reproducing the realism I hear in the real thing and is why I purchased it. To date it is still the sample set I use most. I thought when I first heard Billerbeck that this would be the next one, but after downloading the demo and using it for a few weeks it was back to, nope this is not one I would purchase. It's good but is lacking in that realism and clarity thing again.

Surprisingly, an instrument I heard early on in my Hauptwerk days and it still impresses me to this day is the Schyven / Van Bever organ of Notre-Dame de Laeken Reiner of Pipeloops did long ago.

So, this brings up the next thing..... Because of the size of the instrument as I recall, Reiner made recordings of various registrations for this instrument vs. separate recordings of each individual pipe and I think this is where the lack of realism results from. This is just me thinking and I could be totally wrong, but when each pipe which is a separate recording, and it is put up against other separate pipe recordings and they are all mixed together, I think this is where things go south. Yes, they are sounding together, but they aren't sounding together like they do in the real thing which I'd assume can really throw off the real sound picture. Now, in Reiner's case he chose to sample several stops together at once, which then is an actual recording of what's really going on and hence I think will sound better, take a listen to the sample recordings for yourself here.

http://www.pipeloops.com//product_info. ... ucts_id=30

Of course the drawback to the Schyven / Van Bever organ of Notre-Dame de Laeken is we now have a virtual instrument we can't operate and register like the real thing as we are limited in the individual stops we can choose and mix as we like and instead have to go with the registrations provided. But again, given this drawback just take a listen to the sound, and this is an instrument from 10+ years ago. How much better could it be now? I think it still rivals many of the newest instruments for what it is. Maybe one of the sample set producers has a new approach or work around now for recording several stops at once but still making a set that acts like the real thing?

Marc
Offline

larason2

Member

  • Posts: 752
  • Joined: Thu Feb 04, 2016 9:32 pm

Re: When will we reach this level?

PostTue Jul 19, 2022 10:56 am

Marc, I absolutely agree with you, as I stated before. In this case the recording of the registers together is greater than the sum of its parts. This makes a difference in the sound of the pipes, but also primarily the mixtures, whose relative wind deficient state makes a huge difference to the sound. If sample set makers recorded just a few combinations, like the plenum plus mixtures, and then subbed them in automatically when registered, it would make a big difference. Of course, to do this though, they would have to go back and re-record everything! Maybe a future sample set could incorporate this, but I’m not holding my breath!

I should also add that I think the reason I don't like the idea of increased pipe delay is that in the context of a home setup, it's indistinguishable from latency, and latency always bothers me, even if it is a few ranks compared to others. In a real pipe organ, for some reason it just doesn't bother me as much.
Offline

josq

Member

  • Posts: 912
  • Joined: Tue Jan 24, 2012 12:11 pm

Re: When will we reach this level?

PostTue Jul 19, 2022 12:55 pm

Just to underscore that the sound of an organ is more than the sum of the sound of individual pipes...

On a windchest, the pipes of reeds and mixtures are often placed closest to the valves, and the principal 8' pipes furthest. When multiple ranks on the same windchest are engaged, such a principal 8' pipe will receive a smaller part of the available wind than the other pipes. I think this behaviour can be simulated in the Hauptwerk wind model, but I'm not sure whether this has been done yet by any sample set maker.
Offline

1961TC4ME

Member

  • Posts: 3144
  • Joined: Mon Feb 07, 2011 1:45 pm
  • Location: Lake Minnetonka, Minnesota

Re: When will we reach this level?

PostTue Jul 19, 2022 5:05 pm

So, here's a thought, and maybe I'm all wet here but this crossed my mind. In a real instrument we have a pipe or combination of pipes, or a multitude of pipes, regardless of how you cut it. They're playing alone or at the same time into the same single acoustic. In sample sets we have quite the opposite far as I'm concerned. We have a pipe recorded in an acoustic we can play by itself (which should sound good), OR we can add as many pipes on top of it as we like. The issue to me here then is we have a pipe recorded and being played back in it's acoustic, added by another pipe playing in it's separate acoustic, and another, and another, and so on, I think you get the idea. We're piling one separate acoustic on top of another, and in some cases many many times depending on the number of stops the instrument has. This is quite different from the real thing.

So, how about this idea? We either start with a completely dry set, or truncate the wet set to zero out the acoustic it came with, and instead add reverb through one of the IR's which Hauptwerk now offers in v5 and above? Are we not then playing all pipes into the same single acoustic? Will this add realism? It would be a good test.

I haven't tried this idea but I'm sure thinking I will just to see what the results might be. That or maybe someone with some time on their hands can do a comparison of a wet instrument 'as is' and one that's received the truncating treatment and try it out with one of the IR's in Hauptwerk and let us know how it worked out. Yeah, we're using an IR and that's not perfect either, but is it better?

Marc
Offline

larason2

Member

  • Posts: 752
  • Joined: Thu Feb 04, 2016 9:32 pm

Re: When will we reach this level?

PostTue Jul 19, 2022 6:25 pm

Actually, I believe that usually the principal pipes are closest to the valve, flutes next, and reeds and mixtures the furthest away. But I’m sure others will be able to confirm this too. I’ve used dry sets with IR, like the silver octopus sets. They sound pretty good, the main disadvantage being that they aren’t perfectly dry (so there’s some remnant of the original acoustic), and they still need a fair bit of voicing to sound right “in the space” of the IR.
Offline
User avatar

mdyde

Moderator

  • Posts: 15445
  • Joined: Fri Mar 14, 2003 1:19 pm
  • Location: UK

Re: When will we reach this level?

PostWed Jul 20, 2022 3:58 am

1961TC4ME wrote:We have a pipe recorded in an acoustic we can play by itself (which should sound good), OR we can add as many pipes on top of it as we like. The issue to me here then is we have a pipe recorded and being played back in it's acoustic, added by another pipe playing in it's separate acoustic, and another, and another, and so on, I think you get the idea. We're piling one separate acoustic on top of another, and in some cases many many times depending on the number of stops the instrument has. This is quite different from the real thing.


Hello Marc,

Actually that's exactly how real organs/acoustics behave too. A church/room acoustic is (almost) perfectly a 'linear time-invariant system' in maths/signal-processing terms. Each real pipe effectively has its own separate acoustic (which is slightly different due to its slightly different position is space, relative to the building and listener), and those acoustics add together linearly. Mathematically/acoustically it *is* correct to add (static) wet samples together, whereas mixing dry samples together and then adding reverb isn't, since the latter effectively places them all at the same point in the virtual listening space. In the latter case the loss of spatial information results in a perceived loss of clarity, sounding rather akin to traditional digital organs. If using dry samples + IRs, in order to maintain the spatial information that an appropriately-recorded wet sample set can achieve you would need to use a separate (slightly different) IR for each pipe (so as to position each pipe at a slightly different point in the virtual acoustic). Hauptwerk v5+ allows 1024 simultaneous independent IRs and pipe routing to allow that to be approached if desired. The various 'wet vs. dry' topics on this forum have discussed that at length over the years.

Also:

- A wet sample set needs to have been recorded with microphones kept in fixed positions, and without microphones/perspectives being mixed together (when recording or subsequently), and with initial pipe speech relative delays preserved in order to retain the full spatial cues, and corresponding perceived clarity, that listening to a good CD recording of a real organ can achieve. (Initial relative pipe delays could be preserved within samples if the real organ was played automatically, e.g. via MIDI if so equipped, or via one of those solenoid key-pressing sampling robot devices. I don't know how many existing sample sets have done that, and we do also have enhancement requests logged for adding voicing/producer per-pipe delay adjustments.)

- Hauptwerk v7 has significantly (and noticeably) lower distortion than v5 or earlier, which helps with perceived clarity.

(I'm afraid I don't really have time myself to be involved further in this discussion though.)
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.
Next

Return to Hauptwerk instruments

Who is online

Users browsing this forum: No registered users and 6 guests