Organ matters - Organs matter!

http://www.mander-organs.com/discussion/index.php?showtopic=3008&view=findpost&p=55866
http://www.ibo.co.uk/IBO2005/services/redundant/manResult.asp?manuals=two&Submit=View

"Anyone who wanted a first-rate organ to replace something inferior should look at this instrument very seriously. Worryingly, the IBO page notes that it must be removed very soon."

Paul

Quote from: David Pinnegar on November 10, 2010, 04:40:13 AMwe hope that you might enjoy
http://www.youtube.com/watch?v=RD9FFPhyRd4

I have a recording of Francis Jackson playing this on the Grove Organ at Tewkesbury (the reopening recital when it was restored).  The whole recital is quite something - maybe I'll look into making it available to all...

Paul

Sometimes the first hint is too late:

http://www.mander-organs.com/discussion/index.php?showtopic=2972&hl=

:'(

I've looked around a bit, and found some references to changes in pitch perception with volume.  (Some references also seem to be tying the effect to changing volume, but I've not read the primary documents yet.)  Most of the references are to the work of a researcher called Thomas Rossing.  Just to be clear, there is no physical effect - this is a perceptual illusion.

In summary, a tone with exponentially increasing volume will be heard as rising if it is above 2kHz, and falling if it is below this, the amount of rise or fall depending on the distance of the frequency from the central one (sometimes called "Stevens's Rule").

These tests have been done with pure tones; in the case of complex, musical, sounds, the effect is less clear.  It appears that a low note with lots of harmonics will rise, because the bulk of the energy is above 2kHz, for instance.  Parkin (in "Pitch Change during Reverberant Decay", J Sound Vib. 32, 530 [1974]) noted the sharpening effect after the release of a final organ chord in a reverberant building, which is where this thread started.  I also recall a discussion (which I can't find) of choirs singing flatter in some buildings, presumably because of the reverberation in those buildings having a spectrum that causes a perception of flattening.

You can easily hear the change in pitch with volume by striking a tuning fork hard and holding it very close to your ear.

Paul

In the sample clip, there is no change of pitch - I've just been analysing the spectrum.

I have a nagging recollection that there might be a real effect that can cause pitch change in reverberation (which I believe I've heard, too), but I can't at present find any reference to it.  I'll keep looking...

Paul

That Yamaha speaker (I recall it being shaped like a grand piano soundboard) had a very short life, and understandably dreadful reviews.  There is a clear distinction between reproducing a recording of a sound, and creating a sound through modelling or suchlike.

I know little about the Conn speakers - but the ones I know of were taken out of use after the effect of using them for reproducing a Hauptwerk organ was heard.

Paul

Busy weekend as well, and my broadband was broken before it...

Quote from: David Pinnegar on October 12, 2010, 12:40:06 AMPerhaps the only true recordings of pipes are going to be stereo recordings with a vertically mounted pair directed at both ends of the pipe?

I've never investigated the physics of flue pipes - though I do know that it is not well researched, and that there is some disagreement about the niceties of how they work.  The open end is indeed out of phase with the mouth, but I suspect that is unimportant because the mouth is where the bulk of the noise comes from.  You can consider the mouth of a pipe as an amplifier, in which the resonance in the body of the pipe controls the airflow at the mouth.

I wouldn't at this stage bother with trying to record the two ends of the pipe.  The time to do that will be when we have reached a point at which we find that the reproduction of open pipes is discernibly inferior to that of stopped pipes.

On the matter of speaker defects, my objection to using them deliberately as part of the character of the sound is partly to do with its being unnecessary, and partly because of uncontrollability.  Whatever character is being added to the sound by the speaker, we can add more repeatably in the electronics; and if the character being added is part of the desired sound, why was it not in our recording of the source?  You don't want to add the same character twice, because you then create a different sound - just as a clarinet stop played with or without the addition of the cornet separé is two different sounds.

If you are modelling a sound, there may be situations where using the speaker as part of the model would be possible, or even necessary - for instance, if the speaker was able to simulate the spacial distribution of the sound from a pipe - but I get no sense that we are close to doing that in a realistic way.

Paul

Yes, busy busy busy right now.  I'll try to find time for a couple of posts this evening - otherwise it'll be Thursday...  Certainly I have some comments to make  ;)

Paul

Quote from: David Pinnegar on October 08, 2010, 06:17:58 AM

QuoteIn particular, it leads to a requirement for minimal additional contribution from the listening room, and so, as well as making this rather dead, we design speakers so that the direct radiation in the direction of the listener has a flat response; other directions can go hang, so long as they do not have such prominent peaks or troughs

That is a rather interesting observation and it has been confirmed for decades since pre-quadraphonic attempts at simulating the rear wave for which experiments it was known that the back speakers needed not to be of critical quality. This has particular relevance for people engaged in seriously multi-channel surround sound . . . in which possibly the fashion may be to have all the speakers, including the rear, exactly the same and thereby unnecessarily expensive?

My remarks were about the off-axis radiation of the front speakers; to the extent that the listening room is dry, it doesn't matter what it is like - but listening rooms are not completely dry, so of course the better designers do not ignore it. 

In addition there is a difference between "surround sound" as commonly practised, in which some woofly noises are sent to the rear speakers to give some semblance of envelopment,  and using rear speakers to reproduce the sounds from that direction as accurately as the sounds from the front are reproduced, which is what I do - this naturally calls for identical speakers all round.

Quote

QuoteThe response of a pseudo-omnidirectional speaker may also not be flat in any specific direction (as I suspect David has also observed)

Again I think this works to advantage of realism.

We should not be using defects in the speaker to add their own character to the sound; this character should be in the original sample (or simulation if the sound is being modelled).

Quote... the Doppler effect changing the pitch of whatever else emerges. So when one has any note up to about 150Hz at a high level, any other notes being forced through that woofer rather than the static tweeter are going to be shaken around, backwards and forwards, so changing frequency on every cycle of the low note. For this reason the unit will be unaccountably muddy and not sound natural.

In fact this is not Doppler effect, but phase modulation (but as they are related, this is nit-picking!  In the same vein, it is incorrect to call this a distortion because it is a linear effect, even though undesired).  More to the point, there is a lot of waffle about it, and precious few calculations and measurements.  In fact, at hi-fi levels it is pretty much insignificant.  It is probably largely insignificant for organs too, because the high levels of very low frequencies are generated by very large woofer cones which (a) are probably moving no faster than the smaller cones of petite speakers which cut off higher, and (b) will simply not be capable of reproducing frequencies high enough to be noticeably affected, and so won't be called on to do so.

QuoteThe most natural sounding speakers will be ones where designers have made special efforts to minimise transducer movement

This leads on to the matter of multi-driver systems and crossovers. 

There have been single-cone loudspeakers that professed to cover the full range of hearing, but usually this was actually a rather truncated range; there are speakers with multiple cones attached to one voice-coil - which in effect use a mechanical crossover, which can only be less well defined than an electrical one.  In general, multiple drivers are used to cover the full range satisfactorily.  There are various problems that arise as a result; good design can minimise these for hi-fi and monitoring use both of which require good on-axis performance more than anything else.  However, when considering the off-axis performance, problems appear.  For any driver on a baffle of a certain size, the output will become more directional as the frequency gets higher; there may also be a substantial change in directivity at the point of crossing over.  Also, below a certain frequency the output drops because of an effect known as the baffle step (this can be corrected in the crossover, of course).  All this is already making our aim of an omnidirectional speaker a bit of a nonsense - but there's more!  If the drivers are not perfectly coincidental (as in Tannoy dual-concentric drivers, for instance), there will be comb filtering effects as we move off the axis of the speaker.  Actually, we're quite lucky that our hearing is tolerant enough to let us get away with all this at all.

Another technique that is used to minimise speaker cone movement is reflex loading.  If not well-designed, this can have dreadful effects, but well done* it is a perfectly acceptable technique to reduce cone movement around resonance.

I am not a speaker designer (and I have to go and do other things right now, and will be out this evening), but this should give a taste of the considerations involved.  I may come back with more tomorrow.

Paul

* This BBC report describes the design of the monitor speakers which I use at home; I have six of them...

This topic was being discussed in a couple of threads on the Hauptwerk forum until David was banned there.  At that point I had just made a long post as the result of some careful thought going back to first principles; I am copying that post here (essentially unedited) because I would be interested in David's reaction to it, and to give him the opportunity to continue the discussion a little.  In summary, for those who don't want to read it all, I believe that David's criticism of the use of many types of speaker in organ installations is justified, but that he gives the wrong reasons for it, and this both colours people's reactions to his criticism and fails to direct him towards a clear explanation of his solutions.

Quote from: pwhodges on the Hauptwerk forumFirst, a story. When I was a student, in the 1960s, some friends of mine who are now famous (as far as specialists in arcane areas of audio can be) for their work in (among other things) psychoacoustics, did some investigations into the realism attained using different stereo miking techniques. They played differently miked test recordings of the same performance to various people. The interesting result was that people who knew little about sound reproduction said that they thought one performance was better than the other - because they could hear a difference, and couldn't think of another way to express it.

I think that David is falling into the same trap, and as a result he ends up describing things wrongly even though he starts from the right point. This mismatch of perception is what has been annoying me, and I'd like to put it right.

I work (in my audio hobby, that is) generally in the areas of recording and reproduction - in 3D as it happens, using ambisonics. My aim in reproducing a recording is to play it back in such a way as to reconstruct the original sound field that the microphone has captured in such a way as to encompass the listener's head. This (though more exacting) is pretty much what most people playing back stereo recordings are aiming for as well. In particular, it leads to a requirement for minimal additional contribution from the listening room, and so, as well as making this rather dead, we design speakers so that the direct radiation in the direction of the listener has a flat response; other directions can go hang, so long as they do not have such prominent peaks or troughs that the (minimal) reverberation in the listening room becomes coloured. This leads to the typical design of a hi-fi speaker or near-field monitor (they are not fundamentally different) - and is not anything to do with being designed for pop music or anything like that, which is what David has been saying that irritated me so much!

PA systems have to deal with trying to get clarity across in places which inevitably have more reverberation than a typical domestic setting. They do this by going for directionality as far as possible, to minimise the excitation of the reverberation, and to gain the greatest signal to reverberation ratio at the listener's head that they can manage. The requirements are in fact similar to those above, but the space available allows the use of more extreme techniques, such as line sources, to increase directionality beyond what can be managed from a single box.

Now, if we consider the case of playing the sound of a pipe, recorded dry, in an auditorium, the requirements are quite different - we want to excite the acoustic, and we want to do it similarly to how a pipe would do it, that is, by radiating sound somewhat uniformly in all directions in order to get a similar build-up of reverberation. (The output of a pipe, or any other instrument, is decidedly not uniform in all directions - but attempting to model that non-uniformity is a lost cause with currently conceivable technology, even if a speaker per pipe was provided.) So we are looking, ideally, for speakers that radiate omnidirectionally, which is clearly a requirement deliberately not aimed for by hi-fi speakers, monitors, or PA systems - so David has good reason for saying that those are unsuitable, though the reason is not that which he gives (see story above). The response of a pseudo-omnidirectional speaker may also not be flat in any specific direction (as I suspect David has also observed), because what we would like to be flat is actually the power response integrated over all directions (parallel compromises can be seen in omnidirectional microphones designed for use in the diffuse field). Omnidirectional speakers are extremely difficult to make, because any sensibly sized speaker is large enough to act as a baffle at the higher frequencies (at 10kHz, an obstacle of only an inch or two's width is already significant); but there are techniques which enable one to get a bit closer to the ideal, and I have no doubt that David is using some of them.

I just think it would make life easier if he would (a) recognise what he is aiming for in engineering terms, so that he can use language that actually addresses the issues, and (b) tell us what he is doing, because once the issues are recognised, it will be seen that there is nothing magic about it - just solutions to a defined engineering problem.

Paul