Wood or metal end? Test results...

[ttonon]

Hi Chris, first let me add a comment to my original post, which I was headed to the computer for, but you had already posted. Last night, I was too weary to put all this in. Concerning the technical setup, I should’ve mentioned that, since you are conducting a controlled experiment, using the same microphone in as much the same way as you can, the requirements on the microphone become much less stringent. I do think this is a worthwhile experiment. I also should have added that the volume of your playing of the notes, as measured in the sound file, is very much the same for all notes, and since our hearing responds in some kind of logarithmic fashion, that part of the experiment was fairly well controlled. The tuning of these notes, by the way, is very good, though that information doesn’t show on the charts I posted. I tried to post the Excel spreadsheet I used, but the site doesn't allow uploading that kind of file. Go figure.

 

The debate on how much end plate material affects sound reminds me of the one debating how much flute body materials also affect sound. For many years, in the past, this was an intense debate, though my understanding now is that, among the people who study this both experimentally and theoretically, there’s no discernable difference due to the materials themselves. (I don’t however think the debate has entirely ended, and I doubt it will be.) There can, however, be significant tonal differences because of the different geometries that come about in construction. For instance, metal wall thicknesses can be much thinner, and so the length of side penetrations can differ. I haven’t examined endplates in detail, with measurements, concerning what differences there are in metal and wood construction, but my own current belief is that any tonal differences that do exist are due to the different ways they are made. Think about it. Imagine a wood end plate with the same exact geometry as a metal one. Can you think of any possible way the sound would be different? I can’t. We can of course throw out ideas, leading theoreticians to scramble for explanations. For instance, you are left now with the acoustic affect of the surface itself, wood being porous, and metal being acoustically smooth. Such a difference can show up only as reflected sound waves, and only those sound waves with wavelengths as small as the pore size of the wood would be affected, leading us to conclude that only ultrasonic frequencies would be affected. There’s also a possibility that reflected sound energy would cause more vibration in one material than another. Without analyzing this, my guess is that the sound pressure levels would be too small, but I could be wrong.

 

I do, however, consider the possibility that you agree with all that, and that, in essence you are wondering just what differences in geometry would lead to the often claimed tonal differences. This is an interesting question. In order to answer this, one would have to start simply by making detailed measurements, and then go about the horrendous analysis. Considering the many different open hole geometries, such measurements might seem too difficult, but such features can be approximated by simplifications, such as total open area ratio, etc. But then you might have to consider effects such as unsymmetrical placement of the open hole pattern, the effect of the musicians hand, blocking some of the holes, etc., and so, the scramble continues.

 

Since we are allowed to volunteer our own subjective impressions, mine are that such a debate is a bit of a waste of time, or perhaps only nerdy. As can be seen from the experiment, people, en masse, can often bark up the wrong tree, when convictions rule over facts. I do accept that in a specific instance or so, a given person can actually discern an acoustic difference, but my first guess is that there’s also significant difference in geometry, but more likely, there are other features of the instrument that are really the source for the difference. You have done here a practical experiment with an effort for control, though, as I think you realize, it probably won’t end the debate.

 

When I first listened to your sound file, I did have the impression that the first sound of the first note was marginally brighter than the second sound, and the spectrum seems to verify that. But honestly, the difference, to me, was useless, in practical, musical terms, yet the spectrum does show a decided difference in response at the higher partials. Are we to take this to mean that the spectral approach may be more sensitive than my hearing preference? People can think about this, and maybe an interesting outcome of your experiment is to evaluate how sensitive the Fourier approach really is.

re-reading your post I realise this sentence encapsulates a thought I was trying to formulate. Bright sounds suggest high partials, but could they be a relationship between lower partials.

No, I think from all we know about acoustics and the workings of our own hearing, which cannot violate laws of physics, bright sounds will be the result of the degree to which higher harmonics contribute to the total sound, especially in relationship to the lower harmonics. I can’t see a way that a bright sound would arise because of some kind of interlacing of frequencies over the entire spectrum.

What about partials a third apart which are disproportionately loud?

There are no such things as partials a third apart, over the entire spectrum, but only in a happenstance way. Partials (greater than the fundamental) are also called overtones. With a steady tone of the free reed, overtones are harmonics, which means they are exact whole number ratios of the fundamental. Thus, it’s possible that a harmonic has a “third” relationship to some harmonic that appears at a lower frequency, but this is very seldom, making up a very small portion of the entire spectrum. You might alter your inquiry by wondering if odd harmonics, or every third harmonic, etc. might be the source. For odd harmonics, we have the example of the clarinet, which does emphasize odd harmonics, and we know the characteristic sound of that instrument. For these other kinds of speculations on the spectrum, if the suggested response occurs over the entire range, without predominantly affecting higher partials, it shouldn’t affect the brightness. Now of course the descriptor “bright” is a subjective concept, and I think here we are beginning to deconstruct it. But that’s okay, I’d still make the claim that a brighter sound has more higher partials contribution to the spectrum than a less bright sound.

[chris]

So, is it all down to perception?

Does one hear a difference if 'everyone' believes that there is a difference?

on a slightly different subject (but probably related) I have a Wheatstone brass reeded baritone that had no baffle fitted (was available) when I got it. I found that a couple of reeds at the mid range were a bit 'raucous' and in an attempt to calm it down I refitted the pine baffle - result no 'raucous' note and everything became more ' even'

It didn't even seem to make much difference to the volume (the reason the baffles were removed)

chris (who knows absolutely nothing about acoustics!!)

[ttonon]

It seems to me that you identified the fundamental and then plotted the spectral amplitudes at the exact harmonics of this frequency.

I'm under the impression that inharmonicities between the upper partials are an important part of an instrument's timbre.

I don't expect that it invalidates any of the conclusions that have been drawn so far, but is there the possibility your plots

don't reflect the maximal values of the peaks, since they may be slightly offset from their ideal frequencies?

I'd also believe it if someone could confirm that free-reeds have particlarily harmonic overtones, which would be consistent

with my understanding of how the sound is produced, but I'm not 100% about this.

Hi Dave, you ask good questions, and they can help fix some basic ideas. Part of my answer to your questions appears at the end of my previous post, so check that out. Here, I add that sustained musical tones, whether from a free reed instrument or a woodwind, horn, etc., must be made of periodically occurring pressure pulses. The tone generator of the free reed is a vibrating bar, which vibrates periodically, at its fundamental. There are sometimes contributions from the next higher mode, vibrating at about 6.3 times the fundamental (not an integer multiple), but not at normal playing volumes, and even then, very little. You would have to look very hard to see evidence of this mode in the sound spectra. As for people who claim they can hear it, well, let's simply continue. Thus the pressure pulses from the reed occur strictly periodically, which means that all overtones must be a multiple ratio of the fundamental, otherwise they could not repeat their periodic cycle in unison, and when the fundamental does, resulting in noise, or perhaps, a sound akin to a bell. The same applies to woodwinds and to the sustained notes of a violin. Now, I would like to introduce a bit of noise into the sound of the free reed, because its harmonicity restricts the expression of individual notes, when played along with others.

 

The inharmonicity and tonality you involves musical instruments that operate in a transient, as in struck or plucked instruments, such as the piano string. There is of course a transient when the free reed starts speaking, which involves more than one mode of bar vibration, but this lasts only milliseconds.

 

The overtones I show in the plots are indeed the maximums that occur in the musical tone. Your question is a very good one, because the peaks of these overtones are very sharp, and if you don’t pick their amplitude at precisely the right frequency, the amplitude you calculate will be very far off. I was particularly concerned about this, and in order to satisfy my own mind, I had to delve into just how Audacity works. With this program, they allow you to export the data, but I found that inadequate. I used the highest resolution in calculating the Fast Fourier Transform, which gave me a data point at about every 2.6 Hz of frequency. But if you use only the discreet values from the exported data, you could miss the peaks by as much as 2.6/2 = 1.3 Hz. I thought that would introduce too much error, as remarkable as that may sound, and here I found a saving feature of Audacity. When you use the graphed data, moving a cursor over the spectral plot, onto an overtone, the program curve fits the discreet data points and calculates the peak, spitting out the corresponding frequency at which it occurs. There is of course some error in that a curve fit is a mathematical device, but this error is quite small and acceptable. So my procedure was the opposite of what you suspected: I (Audacity) found the overtone peak first, then got the frequency at which it occurred.

 

Thus, I can assure you that what you see are true peaks, as can be reasonably measured/calculated. In addition - and here's the final confirmation - I calculated the ratio of overtone-to-fundamental frequency for all the notes, and provide the data for you in the attached file. It’s remarkable how accurately these ratios are indeed whole number ratios, and thus, they very strongly indicate that the values shown in my plots are indeed the true overtone peaks. They also confirm my statements, above, and the precision of the Audacity program.

 

Best regards,

Tom

Chris Ghent Experiment, overtones.doc

Edited November 16, 2012 by ttonon

[Dirge]

I was turning this strange business over in my mind. I got to thinking that wood ended instruments sound 'sort of brown like the instrument' and it struck me that I routinely describe the metal ended wheatstone aeola tone as 'silvery' anyway.

 

Perhaps this is where the difference lies; we see the instrument and our brains adds a twist when interpreting the sound signals they are given to work with.

 

If this is so I'm creating a percieved difference in my head. Is this because someone told me I should expect a difference? It will be interesting to see how things sound next time I'm out with a number of players, now this has been suggested. Will the effect vanish?

 

There's a doctorate there for someone...

[DaveM]

ttonon,

Thanks for the detailed reply and additional info.

 

Just speculating here, peoples' perceptions could also affect things in how they play,

i.e. subconsciously modifying their manner of playing to make a metal ended 'tina sound brighter.

[Chris Ghent]

I was turning this strange business over in my mind. I got to thinking that wood ended instruments sound 'sort of brown like the instrument' and it struck me that I routinely describe the metal ended wheatstone aeola tone as 'silvery' anyway.

 

Interesting, I have the same colour/tone association.

[cjmiller]

ttonon,

Thanks for the detailed reply and additional info.

 

Just speculating here, peoples' perceptions could also affect things in how they play,

i.e. subconsciously modifying their manner of playing to make a metal ended 'tina sound brighter.

 

 

From a scientific standpoint, I think you've hit on the weak spot of the test.

 

We didn't know which end was which, but Chris did. As the player did HE subconsciously play them differently?

Did he back off a bit when playing the metal ends, or did he lean into it just a touch more because it had wood ends?

 

I think a truly unbiased and scientific test would require the player to be unaware of which instrument he was playing. No idea how you would do that, but like Dirge said, there's a doctorate waiting for someone.

[ttonon]

From a scientific standpoint, I think you've hit on the weak spot of the test.

 

Hi cj, I don't think so. I mentioned in my second post:

 

I also should have added that the volume of your playing of the notes, as measured in the sound file, is very much the same for all notes, and since our hearing responds in some kind of logarithmic fashion, that part of the experiment was fairly well controlled.

Assuming Chris oriented the instrument in the same way with respect to the microphone, there are only two variables left in the way you play it: choice of notes and bellows pressure, and from visual observation of the sound file, it looks like he did this well. A professional classical pianist once told me that there are ways you could press on the piano keys that will impart more effect to the note than simply the strength of the blow on the strings, alluding to some kind of after touch effect, but of course, he could offer no physical mechanism for it. When it comes to musical perception, some people get religion.

 

Best regards,

Tom

[Randy Stein]

To remake a quote by Groucho Marx..."Who you going to believe, me or your lying ears?"

 

rss

[Chris Ghent]

Let me give you a few subjective details about the test.

 

When I started making wooden ended concertinas some time ago and they were less bright than the metal ended ones I had been making, I thought, the wooden end is less bright, therefore its confirmed, metal = bright and wooden = warm. This was despite the fact I had made a number of other changes at the same time, a clue to my non-scientific thinking at the time.

 

So when I first played the concertina with the metal end fitted I was bracing myself slightly for the assault on my ears and for the wallpaper to roll peeling off the walls. To my complete surprise it was not the slightest piercing. I blamed myself and tried to find its piercing point by playing louder. It just sounded louder. I did try imagining there might be some form of attack that might catch the reed in a strange flexing pattern that might sound brighter, but really there is only extremely fast attack to try and that took seconds. I was left to conclude the tone I was hearing was intrinsic to this metal end on this concertina and it was slightly subdued when compared to the wooden end of the same pattern.

 

I recorded the scale on metal end and the wooden end. I made sure to keep the microphone at the same distance and orientation to the concertina, and I tried to exert the same pressure on the bellows with each note. I edited the two tracks, wood and .65mm, together, alternating the notes. I then did the one thing that might mean it is a little less valuable to Tom, I aligned the volume of the individual notes. These were small adjustments and there was no rationale I could see behind which notes were high or low. The notes from the metal end were random slightly up and down in volume, and the notes from the wooden end were too. Short of building a machine to apply identical force to the bellows I don't see how to standardise the volume.

 

I am confident the recordings reflect what I hear in the flesh.

 

I don't think the bright metal ended concertinas I have heard were not bright.

 

The brightest concertina I have ever heard was a 1934 Wheatstone Linota with metal ends, although come to think of it a friend has an all metal concertina with piercings around the action box sides that can clear a room fairly quickly. However I do think some wooden ended concertinas are bright also.

 

What do others think, have you heard a wooden ended concertina that would drive you out of the room?

[ttonon]

Hi Chris, the differences in playing volume that you mentioned would have nil affect on the normalized sound spectrum, which I plotted. As I mentioned, normalization cancels out volume differences. The only possibility left is whether such volume differences could appreciably alter the spectrum, and I seriously doubt that, for the steady tones you recorded.

 

I am confident the recordings reflect what I hear in the flesh.

Hmmm… might we define here the concepts of flesh hearing and electronic hearing?

 

Best regards,

Tom

[Henrik Müller]

...

Perhaps most importantly, the response of the microphone and amplifying circuit is part of what we hear (and do calculations with).

...

I would add "And so are the surroundings".

 

There is a reason why serious acoustic measurements are done in an anechoic chamber.

What we want is the sound from the object only, not the contribution of reflections from the

walls and any other objects nearby (this will lead to the increase or decrease of certain

frequencies depending on the phase of the reflected signals).

 

A simple test with a real-time spectrum analyzer confirms this: move the instrument around

a little bit and the overtones are jumping up or down. It also shows the dependency of the

pressure - it's near impossible to keep the it constant, this again changes the relationship

between fundamental and individual overtones.

 

So in lack of anechoic chambers in our homes, we'll have to do with our ears -

 

/Henrik

[Chris Ghent]

I would add "And so are the surroundings".

 

Henrik, nice to see you weighing in..!

 

I agree for totally serious sound comparisons elimination of sources of sound contamination need to be excluded, and I don't have the resources for this. Consequently I relied on the old sound man's method, get in close.

 

For my sins I once spent a season teaching television sound techniques, and the example I would use in this instance would be; if you are recording a voice and noise from the next room is at the same level as the voice, if you decrease the distance between the microphone and the subject by half you can turn down the input volume by half and still have the same volume from the subject. The noise in the next room, however, will be reduced by half. If you decrease the distance to the subject by another half and then compensate by reducing the input again the noise in the next room will be reduced to a quarter. I used to demonstrate this rather than say it in this wordy fashion.

 

If the microphone is close to the subject then the echo from walls 3 metres away will be relatively insignificant. In this test I was about 300 mm, 12 inches, from the concertina. I'm not saying there would be no sense of the room, just relatively dininished. I paid more attention to being in the same relative position to the microphone and having the microphone in the same place in the room in both test recordings.

 

Cheers

 

Chris

[Chris Ghent]

Hmmm… might we define here the concepts of flesh hearing and electronic hearing?

 

Tom,

 

this has been on my mind, prompted by the realisation there are things I experience as heard in the flesh but do not hear in recordings. In particular, with a very bright concertina I sometimes experience an uncomfortable effect in the ear which may be a consequence of excess volume but feels like the air pressure in the room has increased.

 

An example, I used to have a conceritna in which the F#5 created this effect, every time it was played I would get an uncomfortable feeling in the ear. When recorded and played back, however, the effect was not present. It may be that it is a volume effect, and recording and playing back being techniques in which volume is necessarily controlled then the effect was eliminated. I remember thinking though, I cannot hear any volume difference in this note compared to the notes around it, but it just does this uncomfortable thing to my ear.

 

If this is a significant effect then a real time spectral analyser, as mentioned by Henrik, might be a better tool than a recording for measuring brightness in a conceritna.

 

Cheers

 

Chris

[Dirge]

From all this analysis of method I deduce that everyone feels the same as me, Chris, that we have huge trouble swallowing the results because it is apparently contrary to a fair amount of personal experience. Hence all the 'did you do this?' stuff. But I looked back and you do say clearly that this is how the 2 sound to you, so it isn't an artifact of the technique.

 

I don't doubt that your experiment was correctly done so I still find it completely baffling. I think perhaps Randy made the best comment...not sure it helps though.

 

And, since you ask, I have never come across a really screaming wood end job, (although it would have to be a Jeffries wouldn't it?) although I have heard some lovely mellow ones, almost sleepy, but I wonder now if they were brass reeded too. The most ear splitting concertina I've heard was an EC with metal ends. "This will make your ears bleed." said the owner with a grin as he got it out of its box.

[Robin Harrison]

And, since you ask, I have never come across a really screaming wood end job, (although it would have to be a Jeffries wouldn't it?)

 

Just to reply to Dirge here.........Paul Read and myself each had a Dipper made as sisters a couple of years ago that are quite the loudest anglos I've heard (and I have heard many Jeffries)...........both wooden ended.

To this point; I've started wearing a soft ear-plug in my right ear after about a year ago I found had a constant pain in my right ear after playing this thing.......I realized that part of the problem was that I was sitting about three feet away from a window on my RHS when practising and I think the very loud sound was being reflected straight back at me....this never happened with a Jeffries (same keys)

I've now moved to a different place but still use an ear plug for this particular Dipper.

I should also note that that at our Toronto English session in a downtown pub, the acoustics are aweful but I can always hear Paul's instrument.

 

Robin

[cjmiller]

All I know is that if my wife is taking a nap, I'm allowed to play my wooden ended Wheatstone, but I'm forbidden to play my metal ended Wheatstone because it wakes her up.

 

That's scientific enough for my purposes.

[Chris Ghent]

From all this analysis of method I deduce that everyone feels the same as me, Chris, that we have huge trouble swallowing the results because it is apparently contrary to a fair amount of personal experience. Hence all the 'did you do this?' stuff. But I looked back and you do say clearly that this is how the 2 sound to you, so it isn't an artifact of the technique.

 

I don't doubt that your experiment was correctly done so I still find it completely baffling. I think perhaps Randy made the best comment...not sure it helps though.

 

And, since you ask, I have never come across a really screaming wood end job, (although it would have to be a Jeffries wouldn't it?) although I have heard some lovely mellow ones, almost sleepy, but I wonder now if they were brass reeded too. The most ear splitting concertina I've heard was an EC with metal ends. "This will make your ears bleed." said the owner with a grin as he got it out of its box.

Dirge,

 

The results were a surprise to me and a perfect "man bites dog" story. You will note I also said "I don't claim this to be a test of all instruments, metal versus wood. I think it shows what happens to this instrument when you fit one or the other end. " This caution to the point of cowardice was both an acknowlegement of the surprise and also of the capacity for error. I have thought long and hard looking for a flaw in the test but have not found one. I would be happy if someone would point out an error, all woud then feel right in the world.

 

I don't know how often this test has done before. I find it hard to imagine the Dippers have not done it. The most important thing is that assumptions are not made from making two "identical" instrument with the only difference being the end material. It has to be the one instrument with two different ends. I suspect the instrument also has to have very good reeds in it. In my experience piercing instruments are good instruments in all other ways.

 

Chris