Organ matters - Organs matter!

Hello all,

This is not an illusion, nor is it a recording artefact.  In the summer I visited Tewkesbury Abbey and, both sitting 1/3 of the way down the north side of the nave and also listening to the voluntary from the crossing, in front of the Milton Organ case, there was a pronounced flattening of pitch during the decay.

At Tewkesbury I could not hear the sound bouncing off the west wall and coming back, so I don't think it's due to that; in fact at York Minster, where I've both sung and played many times, the organ, particularly the 32' Sackbut, is very audible as it rebounds from the back of the nave, but there is no frequency droop.

I've noted this effect in numerous acoustics and it seems to me that it is most pronounced when the width and height of the nave are similar.  York is wide but nowhere near as wide as the 92' of the vault.  Tewkesbury, being Romanesque, is very squat.  My local church, although not very reverberant, also exhibits this phenomenon and is roughly square in cross-section of the nave.

Warm Regards
Neil

Hi All

The problem with religious fundamentalism is that with all ideological systems, the end result is always the same: people who don't agree wind up dead, the prime examples being National Socialism and Communism, neither of which at its source has anything to do with the culmination of both ideologies as they're told in history.  The problem is always the IMPLEMENTATION.  Involve a human and too much of a good thing always winds up the same way.

Religion != Spirituality.  Religion is CONTROL, Spirituality is FREEDOM.  With respect to the members of the Clergy frequenting this site, I don't need someone in a silly hat to tell me right from wrong or act as a spiritual conduit, it's frankly pathetic.  If we're constantly treated like children, how are we supposed to learn and grow? I KNOW that we see through a glass darkly, that this "reality" we inhabit is really conscious co-creation and I don't want to be limited by the closed-minds of those still inhabiting the 7th century.  Oh, you think I'm just talking about the POLITICAL system that is Islam?  Had a look at what's going on in Christian Africa lately?  The Witch Children ring a bell?  It certainly isn't comforting to me to think that all the eye rolling, ullulation and writhing around on the ground casting out demons is probably more like the original Christianity than anything we know in this country.

I've sung in church since the age of 9, was an assistant organist at 16 in a very "high" Anglican church singing choral evensong twice a month but frankly, I tired of everything but the music and the architecture (culture != religion) by the time I was 17 and when I was "outed" as an agnostic by the utterly uncharismatic Vicar who did absolutely nothing to try to help me spiritually, I was more than happy to sling my hook. 

Oh, and I look up from my computer as I hear the dulcit tones of the insufferable idealogue Richard Dawkins.  Why is it not possible for God to have thought to himself one day, "Hmm...Evolution, this could be a lot of fun."  ???

God give me freedom from Religion! (and Atheism's one of them)

Neil

That's how Marshall & Ogletree recorded the samples for Trinity  Wall Street, or so I've read.  However,  I don't think even this is the answer, as summing these two outputs electrically will give much greater phase interactions than the acoustic summing a listener would hear.  And since one can't place one ear at the languid and the other 8 feet (or 16, or whatever) away, surely this is irrelevant?  A lot of peope like spaced-omni recordings of organs which certainly captures the "space" but isn't strictly accurate as a listening perspective; one can't be in two (or more) places at once. 

What about a binaural (Neumann Sphere?) set up near the foot of the pipe but at just enough distance to provide integration between the two sound sources?

Then, didn't I once read on Dr Pykett's pages that very little sound is actually radiated from the open end of a flue pipe?

Best//Neil

Hi Tony

You're right; multiple channels in a real acoustic would be the best option but I'm not in a position to do that.  What is interesting, though, is that I *have* heard the same 4 ranks as in my demo, put through a single pair of speakers in a real acoustic and the mixing problems were exactly the same as in my demo.

Best//Neil

Hi Tony and thanks for listening

To clarify, there are not actually any "speakers" here; the whole thing is "virtual" and the entire signal path was as follows:

"Separate" file: 

Dry samples (Hauptwerk VST) > Reaper (VST Host) > Each rank into a different track > different convolution reverb on each track > Mixdown > HDD recording.

When I say "different" convolution on each track, I don't mean random impulses, I mean the virtual equivalent of placing a starting pistol (or whatever) at the position of C1 and C#1 (C/C# divided chest) for each rank and recording the impulse.    This is the same principle (without speaker frequency aberrations) as giving each rank a separate stereo pair.

In the "Shared" file, the signal path is the same but all the stops were combined electrically BEFORE being passed through just the Open 1's impulses.  This is the same (again, without speaker response aberrations) of putting all 4 ranks through the same speakers.

For me, listening on high-quality headphones, there is no comparison between the two files.  As far as "presence" goes, when I made these files I was only aiming to cure the phase problems, *not* make it sound as if it was a fairly distant recording.  I have moved the goalposts since this proof-of-concept and the end result does, to me, sound thrilling.  Through transducers capable of reproducing it correctly.

Whether or not real speakers are used in a real room or not, isn't relevant; the point of the exercise is to show that electronic mixing of dry stops is detrimental to tutti and nearly all other registrations.

Warm regards//Neil

Dear David

The phasing/lack of richness and blending are best heard, I have to admit, through headphones in this case.  Compared with results on a full division, which no computer at present is capable of rendering with real-time convolution separately on each rank, it is somewhat subtle.  Your reaction, however, is similar to that of the original sample-set producer's in that he commented the rendition of "Offertorium" was "as superb as anything that I have ever heard when it comes to a counterfeit Willis."

I wish I was in a position to hear the "results" through a capable sound-system at realistic levels; I can't approach that through studio monitors in a terraced house, unfortunately.

Presuming that the finished article is as worthy as this early demo from late August, I expect to be "finished" by the middle of next month.

Warm regards
Neil

Gents,

Below please find links to two files I've placed on Dropbox.  Please note that they were both recorded through a VST host in real-time i.e. convolution reverb was applied to the original DRY samples.  This is not a demo of the "finished" new Hauptwerk set as it was a proof-of-concept before I started the lengthy job of actually producing the set. The same midi file was used to play both files and the registration was the same: From the Great, Open 1, Open 2, Open 3 and Claribel Flute.  I had not at this stage done anything to the regulation so please discount this from your observations.

The file "separate" has
each rank routed to its own track, with its own impulse response which mimics where that rank would be on the windchest, with the exact same microphone position, as if one was recording a "real" organ.

http://dl.dropbox.com/u/13017751/Separate.wma

http://dl.dropbox.com/u/13017751/shared.wma

LISTEN TO THE "SEPARATE" FILE FIRST.

The second file, "Shared" is the same material except all ranks are
routed to the Master Output then the impulse response used in the other file for Open 1 is
applied to this mixed output.  In other words, this is what most
people using a dry set and convolution reverb would do.


Notice that in the "shared" file, the ranks DO NOT BLEND. Particularly, the
phase effects are much more pronounced in the held chord at the end of
the file, than in the "separate" file.  In the "separate" file, the
slight phasing you hear is exactly what you would hear if these were
real pipes in a real building and were recorded from the same mic
position.   Consider that all you're hearing here is 4 open 8' ranks.  The effect is worsened the more ranks and pitches are combined on the same output.

NOTHING, not tuning, regulation, voodoo or anything else can
correct the phasing in the "shared" file, other than separating the
ranks.  In the "separate" file, the act of convolving them separately
fulfils the same role as the "air" between a real pipe and a real
microphone, so the output CAN be combined in the same speaker or
output pair.

This is obviously not a real musical test of full organ or even a full division.  I have made a demo of the actual Hauptwerk set using Great to Mixture and then Great to Mixture with reeds 16, 8, 4 to fully prove the concept.  I have linked to that file below. ***PLEASE NOTE*** Since I made this file and produced the Swell, Pedal and Choir, I decided that the reverb was long enough to cure the phasing problem but a longer reverb in the bass would improve the overall effect.  I have not yet produced a demo of the reworked project.  Suffice to say that the increased bass-end reverb (and slightly reduced treble reverb) does not cause the set to be any less playable.

Also note that there is some quite severe beating between ranks; this was intentional as it was intended to prove that the result is an "out of tune organ" it is not the horrible phased sound which occurs on dry ranks when detuning is applied.
http://dl.dropbox.com/u/13017751/WHTPG.wma

I look forward to everyone's constructive comments!

Warm regards
Neil

Quote from: revtonynewnham on September 18, 2010, 03:33:52 PM
One partial solution that I've seen is to reflect the sound from organ speakers off a hard surface, which produces some diffusion, and takes some of the harshness of the tweeters out.  It's an idea worth investigating, especially when a lower than optimum number of speakers has to be used.

This comes back to David's very valid point elsewhere, that loudspeakers do not radiate sound in a manner anything like the radiation from a pipe.  Distance does of course lend its charm (with a real organ), but what I've found is that the tweeter HF fall-off with distance is nowhere near as steep as the fall-off of diapason edge-tone, for instance, which is very perishable.  This is no surprise, since loudspeakers are designed to carry their frequency range over a large distance.  Stand 30 feet back from a speaker replaying the sound of a diapason recorded 3" from the languid and it will sound...much like a diapason recorded 3" from the languid.

It was the realization of this fact, and also that with a fairly-dry set, replaying it in a more reverberant space does not cause Dr Pykett's "ambience conflict" which does occur with a "wet" set in a drier space, which convinced me to work on my current project. 

The method I use (which, yes, I shall be cagey about) uses ONLY diffuse sound.  I do not mix diffuse and dry sound because I have found this does not solve the phase/mixing problems.  Although it is fully diffuse, I have "recorded" from no more than 8m from the source which, for the "room" used, means the mics were actually within the reverb radius of the case-front pipes.  The furthest-away pipes were actually the Great and Swell reeds as they were at the back of their respective soundboards.

The crux of the "mixing" problem arises because where pipes mix acoustically, the phase interactions are of a much lower magnitude (peak to trough) than with purely electrical mixing, which is what we have when playing dry stops through the same speakers.  It is for this reason I say that NO loudspeaker can overcome this, because it is a problem of physics.  The Lowther-style "full-range" devices I assume from his ebay pages that David is using will help with the treble beaming problem due to the natural roll-off of the "whizzer" cone, thus improving the tonal balance of the tutti, but not fix the phasing problems.

Another reason I started my project is that, with the exception of the Lavender Audio Haverhill OIC, I have yet to hear or play a "wet" Hauptwerk sample library which was remotely suitable for home practise. I feel that the current crop of "archival" sets is just not suitable for a serious organist.  Either the purchasers raving about how wonderful they are suffer from a serious case of cognitive dissonance (I've paid hundreds of bucks for this, so of course it's fantastic) or are scared to voice an opinion that grates with the party line.  Shades of The Emperor's New Clothes? I've played several cathedral organs with detached consoles and it is not a pleasant or involving experience.  Why anyone would want this for a home instrument is very puzzling.  What Hauptwerk currently offers seems not to be,

"Play historic instruments one could never travel the world to gain access to" but rather,

"Listen to historic instruments playing repertoire not available on CD"

...but that is another topic.

I will link to an A-B demo later today

Warm regards
Neil

Hello all

I refrained from entering this topic on the Hauptwerk forum because there is a general animosity towards dry samples in that arena, considerably but not totally advanced by one sample producer in particular who has been heard to say in public that the only "valid" use for Hauptwerk was with touchscreens, wet samples and near-field monitors or headphones.  I've been told that this has turned several pipe and/or hybrid producers totally off Hauptwerk as they accordingly see it as of "no use" to them. I have for a long time refused to drink this wet-organ "Kool-aid" and maintain that a set recorded from half-way down the nave - however historic or accurate - is of very little use to a serious organist looking for a quality practise instrument for the home.

I digress.  I will try to not make what follows sound like an advert, because it is in no way intended as such.

Martin Dyde has mentioned - and indeed experimented to prove - Colin Pykett's theory of signal mixing, although David's original forum posts seemed to sidestep this issue.

Since Martin's original experiment, I have had the opportunity to hear and play a 16-channel Hauptwerk installation in a somewhat reverberant church, playing a 4-manual "dry" library I have been involved with personally over several years, and a "slightly wet" set which, to my knowledge, did NOT employ release truncation in the building.  Most of the channels were powered by Behringer "Truth" monitors but the Pedal went through a pair of Tannoy Westminster floorstanders and was augmented by two custom-made active subs which were extremely effective on the Double Open Wood.

The results were interesting.  On the "dry" set, the smaller registrations, diapason choruses and even the solo reeds, sounded very good indeed, as did the plethora of string tone available.  However, even employing 8 stereo pairs, as soon as the stops were piled on, the sound became increasingly "electronic" due to the comb-filtering caused by the "signal mixing effect."  Full organ was hindered further by lack of amplifier headroom.

The effect of the "slightly wet" organ was markedly different, bearing out Martin's assertion that fewer channels are needed when one has, effectively, a CD recording played back in real-time.  It was also interesting to note that the fact it was "an acoustic played within an acoustic"  was not obvious at all, probably because the original recording was considerably drier than the live room.

Since my domestic situation (terraced house) precludes the use of my home-built Wilmslow Audio Gemini monitors and Decware "Wicked One" W-horn sub,  I have to play mostly through headphones. 
My experience with dry samples over my 7-year affair with Hauptwerk has convinced me that whilst they are infinitely more suitable for serious practise purposes than the fully "wet" sets, which I view as merely toys, a multi-channel audio system separating, at the very least, the unison stops is not desirable, but obligatory.

In my experience, no amount/quality/sophistication of convolution or algorithmic reverb (on a single stereo output, whether that be for a division or whole organ) will compensate for the blending problems caused by the signal mixing phenomenon.  I contend that no speaker in existence will ameliorate these problems.  My live experience, and the subsequent production of a Hauptwerk organ with a microphone position of 8m from the pipe-fronts, convinced me that the "ideal" set for home use is neither "dry" nor a "wet" set such as Salisbury (which I find unplayable for more than a few minutes), but is in fact the kind of sound one would get standing "reasonably" close (i.e. less than 30ft) to an organ, in a church or concert hall with less than 2.5 seconds reverberation on full organ/loud reeds.

I then set out to produce this, by creating a building artificially and then making stereo impulse responses for every single rank, arrayed in space such as to mimic the windchest layout of a pipe organ.  Each pair of impulses was "recorded" with exactly the same microphone setup whose location with reference to the building layout remained static, as would be the case were I recording a "real" pipe organ. 

Every dry rank was then re-recorded in a VST host using its own impulses, effectively producing an entirely new sample library. Each rank was regulated to the "acoustic" before recording, and recorded with several sets of multiple releases.  Since a MIDI file was playing the notes rather than wetware (me), I know *exactly* when each key was released and all the "full" samples are exactly the same length, which makes processing them a very rapid process compared with producing a "wetware" set.

Whilst it has taken me upwards of 300 hours to produce this set, the results are certainly worth it and far more convincing than I ever could have hoped.  Adding to an 8' chorus, the sound does not just get louder as with fully dry ranks played through the same speakers/headphones, but rather it gets richer and any inadequacies/variances in the individual voices are magically compensated for by the others, as in reality.  The original producer of the samples was somewhat skeptical that my methods were valid, until he heard the results, which he described as, "frankly unbelievable."

To get somewhat back onto the point, whilst David is right that very careful voicing/regulation must be employed, it is my experience that for dry stops played through either a single stereo pair per division, or indeed a single pair for the whole organ, NO amount of voicing or regulation will give a realistic result on a Tutti or anything approaching it, nor will any Franck chorale registration really sound like massed foundation tone.

Before anyone castigates me with something akin to the "put your money where your mouth is" comments David received on the original HW forum threads, I hope it suffices to say that as a result of my efforts, which I hope to complete by mid-November, I am preparing to build a 4-manual console employing dual 23-inch touchscreens, in the style of Randall Mullin's (the Ikea table method) solely to run my new sample set, therefore my own investment in the principles I have outlined above is quite considerable.

In summary, I would like to close with the comment that if one's "dry" Full Swell doesn't sound like St Paul's or St Bees, it may not be the voicing, the regulation or even the choice of speakers, but rather that too few speakers are employed.  What I've done is effectively give each rank its own pair of speakers, placed the whole shebang into a "live" acoustic and produced something with all the blending characteristics of a "wet" set (or CD recording of a real organ) without any of the playability drawbacks.

If anyone is interested in hearing an A-B comparison of 4 8' foundations played through one pair of "speakers" and then four pairs, I'd be very happy to provide it and serious demonstration pieces will follow in due course.

Warm regards
Neil