The Sound of the Western Free Reed

[ttonon]

I think most of us would agree that the Western free reed has a unique sound, and that the general public has a rather complex reaction to it. Although many times I feel the sound appropriate within the context of the music being played, I sometimes come away feeling that there are problems with it. The musical instruments I’m talking about here of course include all the keyed variety, along with the harmonica.

 

I assume that suitable explanations for my reactions can be found in the acoustic sciences - Psychological Acoustics as well as Physical Acoustics – and I usually spend more time pondering the latter. There, I usually focus on the free reed’s sound spectrum, which is rather full, containing many harmonics, both odd and even, numbering into the dozens. Such a spectrum shape is different from all other reed instruments, which are relatively limited in the significant number of harmonics that make up their musical tone. For many notes, these other instruments contain only two or three significant overtones.

 

Could it be that the fact of so many harmonics in the sound spectrum of the free reed is responsible for the discomfort I sometimes feel when I hear its musical result? Thinking further, it’s usually when I hear more than one reed sounding at the same time that this negative reaction occurs. The combined sound sometimes comes out smeared, smashed together. When chords are played, the sound is often a wheeze, wherein individual notes can manage only the hint of presence above the surface of sound, largely drowning below.

 

Many readers here have probably noticed that I’ve left out the important start transient of the musical tone. I’ve done this purposely, not because I don’t think it important, but rather because I think the steady state sound of the reed itself is a worthwhile issue to focus on.

 

Let’s say we have two reeds sounding, one an octave above the other. In this case, and because of their sound spectrums, it’s a wonder that the human auditory system can even discern the fact that two reeds are sounding. It would be very difficult, if even possible, for our eyes to pick out in the combined sound spectrum the contributions made by separate notes. In contrast, consider two clarinets or two trumpets playing the same written notes as before. Because of the drastic reduction in the number of overtones, and the sometime dominance of odd harmonics that quarter-wave instruments provide, one can easily pick out the two separate notes in the combined sound spectrum. I can’t help but think that, if we could find some way to throw away many – even the vast majority – of the harmonics that literally explode out of the sound of the Western free reed, we might have a family of more versatile instruments. By more versatile, I mean the ability to better join with the sounds of other instruments – both of the same kind and otherwise – without becoming muddled in the background soup of harmonics.

 

The stringed instruments also provide a full spectrum, with many harmonics, yet people seem to react very differently to their sounds. But here, we must make an important distinction between the stringed instruments that produce sustained vibrations and those that produce transient vibrations. The bowed instruments produce sustained vibrations, and so, like the free reed, their overtones must be in exact harmonic (whole number) ratio. The struck or plucked stringed instruments produce transient vibrations, and their overtones – because of a variety of complex physical phenomena - are not in exact harmonic ratio. When you play a chord – say a biting jazz chord – on a piano or guitar, you hear a vibrant sound wherein each and every note can be discerned. As many here know, the sound of the plucked string instruments contain stretched overtones, due in part because of the real stiffness in the strings. These stretched overtones are another way of saying that the harmonic structure does not contain exact whole number ratios – or, in other words, there’s a certain amount of healthy noise in their sound. Could it be that the non-harmonic effects within the sound of the piano and guitar allow each note to maintain better its own identity, causing us to enjoy such crisp chords? I suspect this is true.

 

It’s interesting to note also that I sometimes get the same discomfort when hearing more than one bowed string instruments playing together. The bowed string, in providing a musical tone, must produce a spectrum with exact harmonics – just like the free reed. Regrettably, and like the free reed, there thus seems that there’s little room for a bit of healthy noise in their tone. I thus suspect that herein lies more evidence for the liability provided by such a full spectrum of exact harmonics.

 

There isn’t much we can do about the spectrum of the Western free reed. Its production arises from a train of regularly pulsed pressure waves, made in precisely the same way each time. Our brains effectively perform their version of a mathematical operation, called Fourier Transformation, and there’s not much we can do about that either. Every once in a while, I do an experiment on this, but so far, I can suggest no remedy, and I’m left only to wonder how long such an overpopulation of harmonics will dominate us and our beloved free reed instruments. I also wonder what others might think on this topic.

Edited March 5, 2011 by ttonon

[Ardie]

This topic is really a challenge but since I sympathize with what you say I try some preliminary comments:

 

.. the Western free reed has a unique sound, and that the general public has a rather complex reaction to it....I sometimes come away feeling that there are problems with it.

 

I guess that the early routine adding another reed set in either octave or "wet" tuning reflects intentions to deal with the awkwardness that might originate from free reed instruments lacking the 'harmonious' experience habituated by listening to instruments with a natural harmonic spectrum of overtones or with resonators in the construction.

 

.. the free reed’s sound spectrum, which is rather full, containing many harmonics, both odd and even,.. Could it be that the fact of so many harmonics in the sound spectrum of the free reed is responsible for the discomfort I sometimes feel when I hear its musical result?

 

I definitely think so.So many "musical people" not personally engaged in free reed instruments or their music spontaneously say they dislike the "rough" tone of harmonicas, accordions etc.

 

Thinking further, it’s usually when I hear more than one reed sounding at the same time that this negative reaction occurs. The combined sound sometimes comes out smeared, smashed together. When chords are played..,

 

Certainly a greater number of 'disharmonic' interferrences are expected to appear in free reed chords than stringed (say harp etc)chords

 

I can’t help but think that, if we could find some way to throw away many – even the vast majority – of the harmonics that literally explode out of the sound of the Western free reed, we might have a family of more versatile instruments.

 

A daring thought since free reed fans will object energetically but I much agree. We have to exclude and retain the blues harmonica for raw blues but apart from that I am sure free reed instruments might be liked more among the "symphonic" music audience that way.The most "harmonic" concertina for example that I know is the "Clarionet" model and it would be very interesting to compare its tone spectrum with 'normal' concertinas.

They have specially formed reeds, wider at the top, and large volume reed chambers. There have been trials with adding resonance tubes to accordions but the need for too large measures put an end to it. You've got the "Cassotto" constructions as well and with harmoniums there are several variants to modify the tone - so surely some possibilities do exist.

[ttonon]

Thanks Ardie, for the comments.

 

I guess that the early routine adding another reed set in either octave or "wet" tuning reflects intentions to deal with the awkwardness that might originate from free reed instruments lacking the 'harmonious' experience habituated by listening to instruments with a natural harmonic spectrum of overtones or with resonators in the construction.

 

You're probably right. Though, in many ways, this approach can make things worse, adding even more clutter to the tonal spectrum. In a sense, it suggests that free reed instruments have an identity crisis: "Just what am I supposed to sound like?" Hence, on accordion register switches, you find labels such as "bassoon," "clarinet," "organ," "piccolo," etc.

 

We have to exclude and retain the blues harmonica for raw blues but apart from that...

 

But there, you don't often hear the acoustic sound of the harmonica. With Chicago-style blues, the usual method of blues harmonica amplification is to stick a low-fidelity microphone up to it, cupping it with the hands (which also suppresses feed-back). The resulting sound is Big and much more dynamic/potent than the acoustic sound, enabling competition with wailing electric guitars, even. And often in acoustic/delta style blues, accomplished blues harp players will use a mic.

 

The most "harmonic" concertina for example that I know is the "Clarionet" model and it would be very interesting to compare its tone spectrum with 'normal' concertinas.

They have specially formed reeds, wider at the top, and large volume reed chambers.

 

Can you point me to any sound files where I can hear this instrument?

 

There have been trials with adding resonance tubes to accordions but the need for too large measures put an end to it. You've got the "Cassotto" constructions as well and with harmoniums there are several variants to modify the tone - so surely some possibilities do exist.

 

I agree, though the effect on the un-amplified sound is marginal. The cassotto, however, does I think, illustrate my point, because its affect on the tone spectrum is to emphasize a few of the lowest harmonics, making higher harmonics less prominent, resulting in a bassier, more ballsey sound. With the bassoon reed, the cassotto is the most common choice for jazz accordion players.

 

In the end, perhaps the most effective way musicians who don't like the acoustic sound deal with this issue is to modify the amplified sound, with suitable electronic filtering (equalization).

 

Apart from Simon Thoumire, I'm not aware of many jazz concertina players, and at the moment, I can't recall just what Thoumire's sound is like, and whether he has a really acoustic sound or a more electronically modified sound.

 

Best regards,

Tom

[Ardie]

I guess that the early routine adding another reed set in either octave or "wet" tuning reflects intentions to deal with the awkwardness that might originate from free reed instruments lacking the 'harmonious' experience habituated by listening to instruments with a natural harmonic spectrum of overtones or with resonators in the construction.

 

You're probably right. Though, in many ways, this approach can make things worse, adding even more clutter to the tonal spectrum.

 

Maybe we ought to check some sound spectrum to see this better but do we not expect both from the octave doubling and the wet doubling that the fundamental will be relatively amplified vs the partials ? and that would make sound virtually "sweeter" would it not? It seems to me that many listeners get that impression anyway

 

We have to exclude and retain the blues harmonica for raw blues but apart from that...

 

But there, you don't often hear the acoustic sound of the harmonica.

 

Indeed - but IF we want acoustic instruments to do the job at all...And your own invention is meant doing so too is it not?

 

The most "harmonic" concertina for example that I know is the "Clarionet" model and it would be very interesting to compare its tone spectrum with 'normal' concertinas.

They have specially formed reeds, wider at the top, and large volume reed chambers.

 

Can you point me to any sound files where I can hear this instrument?

 

Not instantly but I will give it a try.Maybe some reader got one? There should be some at Hornimans I believe.

 

The cassotto, however, does I think, illustrate my point, because its affect on the tone spectrum is to emphasize a few of the lowest harmonics...

 

I wonder...I have thought that the cassotto effect rather would be that it works 1)as a "formant resonator" similar to the vocal tract i e that it emphasizes a wide part of the spectrum ( below 1000 Hz) and 2) by absorbing/damping amplitudes in the higher part of the spectrum.Or do we mean the same?

This pricniple ought to be possible to practise with concertinas too for example by increasing the volume of the compartment between endplate and action board. I have heard that George Jones actually had experimented with such kind of instruments.

 

 

In the end, perhaps the most effective way musicians who don't like the acoustic sound deal with this issue is to modify the amplified sound, with suitable electronic filtering (equalization).

 

Maybe so - and that will be the end of all acoustic instruments except for nostalgics won't it?...

[ttonon]

Hi Ardie,

Maybe we ought to check some sound spectrum to see this better but do we not expect both from the octave doubling and the wet doubling that the fundamental will be relatively amplified vs the partials ? and that would make sound virtually "sweeter" would it not? It seems to me that many listeners get that impression anyway

Let me figure out what you’re saying. Let’s suppose that the two reeds produce identical spectrum. As I see it, sounding both in the same way and at the same distance from your ear would then simply double all the partial pressures making up each harmonic (pressures add). Since intensity varies as pressure squared, the harmonics that have larger partial pressures in each individual spectrum would maintain the same relationship with respect to intensity. For instance, say Harmonic A has pressure amplitude 1.2 units, and Harmonic B has 1.0 units (a typical variation). Their intensity ratio is the square of (1.2/1), or 1.4 units. With two reeds, the pressure ratio becomes (2.4/2 = 1.2), which is the same, giving an intensity ratio again of 1.4. I thus fail to see how simply playing two identical reeds will change the quality of the tone. With off-tuned reeds, I think the principle still holds. Another fact to consider is that the fundamental in the free reed spectrum does not always dominate. Sometimes it’s the second, or even third: complications arise because of the presence of a cavity and other complex geometric structures.

I wonder...I have thought that the cassotto effect rather would be that it works 1)as a "formant resonator" similar to the vocal tract i e that it emphasizes a wide part of the spectrum ( below 1000 Hz) and 2) by absorbing/damping amplitudes in the higher part of the spectrum.Or do we mean the same?

I myself refer to the cassotto as a “reverberator,” and not a “resonator,” but I admit, the distinction isn’t very clear. The principle here, I believe, is due to the doubling of pressure amplitude within a distance of about a half wavelength when a sound wave bounces off a rigid surface. I think of it as a transient-in passing effect, and not a steady state effect, as with a true resonator, but again, I wouldn’t push the distinction. In any event, the walls of the cassotto provide many surfaces from which the sound can bounce. The overall dimensions of the cassotto are important design parameters, and ideally, the geometry allows satisfactory effect over the entire range of the instrument. But it doesn’t, at least from my own observation: the lower notes/registers are significantly more affected than the higher.

This pricniple ought to be possible to practise with concertinas too for example by increasing the volume of the compartment between endplate and action board. I have heard that George Jones actually had experimented with such kind of instruments.

My guess is that a cassotto-like geometry can be incorporated into concertinas. The small size of the instrument, however, may restrict how effective such devices would be for the lower notes, or may require designs that differ substantially from those found in accordions.

[Chris Ghent]

My guess is that a cassotto-like geometry can be incorporated into concertinas. The small size of the instrument, however, may restrict how effective such devices would be for the lower notes, or may require designs that differ substantially from those found in accordions.

 

Tom,

After reading this I went out this morning and increased the volume of a left hand action box with a 1/4" spacer. This took the volume up by 40%. I can detect no difference in the formant. It doesn't show up to the ear, in a recording before/after, or in the harmonics display either.

 

If 40% makes absolutely no difference then it would need to be a threshold effect, where more than one condition is met. What would the other condition/s be? If it needs to be some proportion of a wavelength then might it need to be different for every reed?

Chris

Edited March 10, 2011 by Chris Ghent

[Ardie]

Maybe we ought to check some sound spectrum to see this better but do we not expect both from the octave doubling and the wet doubling that the fundamental will be relatively amplified vs the partials ? and that would make sound virtually "sweeter" would it not? It seems to me that many listeners get that impression anyway

Let’s suppose that the two reeds produce identical spectrum. As I see it, sounding both in the same way and at the same distance from your ear would then simply double all the partial pressures making up each harmonic (pressures add)....

.. With off-tuned reeds, I think the principle still holds....

 

Tom, you are right, but it isn't the whole picture is it? With "off-tuned reeds" sounding simultaneously - as with two in octave tuning or two in wet tuning - you get the mathematical series of combination frequences which are added to the sound and I assume *this* is what constitutes our perception and the impression that the octave tuning and the wet tuning result in "sweeter" tone.Or may other factors be involved as well?

 

My guess is that a cassotto-like geometry can be incorporated into concertinas. The small size of the instrument, however, may restrict how effective such devices would be for the lower notes, or may require designs that differ substantially from those found in accordions.

 

Tom,

After reading this I went out this morning and increased the volume of a left hand action box with a 1/4" spacer. This took the volume up by 40%. I can detect no difference in the formant. It doesn't show up to the ear, in a recording before/after, or in the harmonics display either.

 

If 40% makes absolutely no difference then it would need to be a threshold effect, where more than one condition is met. What would the other condition/s be? If it needs to be some proportion of a wavelength then might it need to be different for every reed?

Chris

 

Chris, what exactly did you do? How is your "spacer" added? When I tried expanding the space by adding an extra 10mm wide frame between action box and end plate ( and extended the buttons correspondingly) I experienced a tonal difference ( mellower, more sonorous sound) subjectively. I didn't have facilities to objectify it. Could you make a sound spectrogram (is that what do you refer to with "harmonics display") to visualize it?

My very subjective impression when comparing various instruments has been that the size of the "endbox" does play a part for the sound but my primary hypothesis has been that this firstly is caused by absorption/damping of higher frequences and this secondary hypothesis regarding a possible "formant resonance" is a speculation rather.It would definitely be interesting to continue the trials...

[ttonon]

If 40% makes absolutely no difference then it would need to be a threshold effect, where more than one condition is met. What would the other condition/s be? If it needs to be some proportion of a wavelength then might it need to be different for every reed?

Chris

Hi Chris, I chime Ardie’s request to please explain exactly what you did. Where is the ¼-inch dimension? How did you measure that the volume is 40% more? How did you evaluate the spectrum? Etc…

 

Without any other info, I’d volunteer that it would be surprising to me if you could detect no significant change in the sound spectrum with a 40% increase in volume, using some sort of baffling. You’ve made a near perfect amplifier!!

 

One exercise I like to suggest is to take your concertina, stand very close into a corner of a room, or put it into the confining shelves of an empty bookcase and see if you prefer the sound – at least for some notes – over what you hear out in the open, or when you hear the instrument played by someone else. If you prefer the more reverberant sound, then there are grounds to possibly explore how a cassotto (tone chamber) might be worthwhile.

Could you make a sound spectrogram (is that what do you refer to with "harmonics display") to visualize it?

Hi Ardie, there’s free software available that enables you to plot the spectrum (Fourier Transform) of acoustic files. Very convenient is the fact that you can export the data into a text file, from which you can put it on a spread sheet. The name of the program is Audacity: http://audacity.sourceforge.net/

It’s made by those generous open source people who have visions for a better world.

[Chris Ghent]

Tom,

 

what I did was to remove the action box, take it apart and place a spare action frame 6.25mm in thickness between the two halves and then reassembled it. If you are still in some doubt as to what this means look at this photo, it is of the lower half of an action box. The frame it is the rim around the outside, with the black top. I had a spare unattached one, I put it on top of what you see here then put the end back on. This took the internal height inside the action box from 16 to 22.25mm, an increase of around 40%. I wrongly reported this to you as a 40% increase in volume, it is in fact a 40% increase in height. In volume it would be larger than 40% because there is a lot of used space in side there, action, pads, buttons etc. If I am understanding you right then height is the defining parameter, not volume.

 

The recording I made of the two states, with the spacer and without, is attached. Each note is played once in one state and then once in the other.

 

Thanks for the heads up on the ability of Audacity to display the spectrum, works well. I used it to recheck this recording, note by note, it confirmed my results. You may find aspects of the 6th pair spectrum interesting, this is the result of a valve which is too thin.

 

To my reading "Ardie" and I did the same thing (his spacer was thicker), but my results nothing like his. Either one of us is wrong or more factors are at play. I'm hoping you will tell me I have made a mistake!

 

Best Wishes

 

Chris

actionbox mod trial.MP3

Edited March 11, 2011 by Chris Ghent

[Ardie]

what I did was to remove the action box, take it apart and place a spare action frame 6.25mm in thickness between the two halves and then reassembled it. If you are still in some doubt as to what this means look at this...

The recording I made of the two states, with the spacer and without, is attached. Each note is played once in one state and then once in the other.

 

To my reading "Ardie" and I did the same thing (his spacer was thicker), but my results nothing like his. Either one of us is wrong or more factors are at play. I'm hoping you will tell me I have made a mistake!

-

Chris, just trying to make the picture complete...Did you reassemble the whole instrument - ends/extended ends, bellows an all and played the instrument in both versions? Maybe you might add a photo with the completed instrument? I am just a bit curious to see what the entire instrument looks like. One factor of importance is if your endplates are comparably open ones - in that case you expect much less ( if any) effect from any change of this space underneath the endplates.

 

Another problem is that a recording of this kind is a very crude and really not much valid method to illustrate discreet sound variations. As a matter of fact our hearing often is a much better instrument in detecting those subtilities. This is a great problem of course when discussing the experiences at a distance.I could NOT for my part for example notice tone differences comparing note by note as with your recording. The impression I got that there was a significant sound difference ( and which I felt personally sure of was "real") came when playing music. I actually did try making some sound spectrum recordings and graphs but it is long ago, the equipment was not satisfactory, and I found nothing... but that as you understand from what I said above I was not very surprised..

[ttonon]

Hi Chris,

There were a couple misunderstandings there. By “volume,” I thought you meant volume of the musical tone. That’s why I asked you how you measured that and why I was amazed you could reportedly increase that volume by 40% without discernible effect on sound spectrum.

 

Thanks for the figure. It’s my understanding then that you put the cover, or grille on top of that spacer, which should restrict the depth to which you can then push down the keys - by the same distance as the extra spacer thickness. No?

 

So you basically are exploring the effect on the sound spectrum by increasing the space between the action board and grille. My first impression is that the effect of this maneuver would be influenced by a number of things, including, the percent open area ratio of the grille, as well as the position of the grille pattern details in relation to other parts (and distances) within the cavity, the total volume of the cavity, the geometrical orientation of the various mechanical parts within the cavity, several of the key distances within the cavity (grille to action board, chord length between spacer sides, etc. In other words, everything in the cavity. I know, it’s not a trivial determination. One thing I might suggest is that you obtain a copy of some free software that allows you to calculate acoustic modes of rooms and other volumes. I have one of these, and if you’d like, I can upload it. I’m not sure how well this will work, but you could plug in different space dimensions as trial and error, and the program will calculate the frequencies of the acoustic modes. A serious issue, however, is the partially opened grille, which this program I don’t think will accommodate. Perhaps there are programs available that allow as input porous structures. One thing can be said right away is that all of these vibrational modes will have wavelengths that are of the order of the key dimensions in your space, and simply knowing the overall size of your concertina, you can calculate very easily by hand these wavelengths. You will probably conclude that all the amplified frequencies will be around and above several hundred Hertz. This is because solid walls give a zero velocity boundary condition, meaning that half wavelengths will fit between the key dimensions. For instance, for a 8-inch distance, or 8-inch half wavelength, a full wavelength will be 16 inches, corresponding to a frequency of 770 Hz. I’d think that an 8-inch length would be a maximum of the characteristic dimensions in question, and thus all other dimensions will lead to higher frequencies. It’s thus easy to see that any effect of this approach will be to influence the higher harmonics in most of the musical tones.

 

I took the data you provided, loaded it into Audacity, analyzed each sound, generated the sound spectrum, exported it to a text file, entered it into a spread sheet, then massaged the data. The results are attached. You can look at this approach and probably manage to do it yourself when you want. In this spread sheet, you will see the fundamental frequency, as measured, with the name of the note. You should verify that these are correct. The Ordinate on the graphs give a normalized response such that the fundamental will have an amplitude in zero decibels. There are more than one graph more most of the pairs of sounds, with different graphs showing different Octave ranges.

 

My cursory analysis looking at these spectra is that there is indeed only little difference between the control and the modified instrument, and it’s not surprising that they sound the same. However, the graphs do show that there's somewhat of a consistent trend in that the higher harmonics are a tiny bit more prevalent with the second sound of each pair, exceptions being for the higher notes, where there seems to be a toss up. If I had to guess, I’d say that you placed the modified instrument sounds first, with the control sounds second. Please let me know if this is true, because it would then indicate that this Fourier Transform method of analysis may be more sensitive than my (and your) hearing. Take a look at the spectra yourself and see what you think.

 

You gave us an mp3 file, which contains some loss of information. I normally don’t waste time with mp3 files when devoting time to analysis and usually like to work with .wav files, which have much more information content. I can’t right now say how much of the variations we see here are due simply to the effect of the compression in the sound file. I did have to throw away the data for about a top third of the octaves, because it looked like junk. Since, however, there does appear to be a consistent trend for the first 30 or so octaves, perhaps the data is really worth analyzing. If you’d care to take the time, for homework , you can pick off on these plots the harmonics that show significant difference between the two sounds, writing down the frequencies. You might see certain frequencies repeated. If so, you will have an idea of what resonances, or reverberations are occurring in your instrument; i.e., a “fingerprint,” if you will. Then, you can convert these frequencies into a wavelength, then to a half wavelength. You can then make various measurements inside your space and see if you can identify those half wavelength measurements with key dimensions in your space. If all is successful, you might get an intuitive feel for what particular dimensions are more important than others. Good luck.

 

Best regards,

Tom

 

Note: Chris, I'm unable for some reason to upload xls files. At first I thought my error was that the file size was too large, so I broke it down into three separate files, but that didn't work. I'll try sending you the full file by email. Above text discussion is of course still valid.

Edited March 11, 2011 by ttonon

[Chris Ghent]

Thanks for the figure. It’s my understanding then that you put the cover, or grille on top of that spacer, which should restrict the depth to which you can then push down the keys - by the same distance as the extra spacer thickness. No?

 

yes, sorry about the confusion.

 

So you basically are exploring the effect on the sound spectrum by increasing the space between the action board and grille. My first impression is that the effect of this maneuver would be influenced by a number of things, including, the percent open area ratio of the grille, as well as the position of the grille pattern details in relation to other parts (and distances) within the cavity, the total volume of the cavity, the geometrical orientation of the various mechanical parts within the cavity, several of the key distances within the cavity (grille to action board, chord length between spacer sides, etc. In other words, everything in the cavity. I know, it’s not a trivial determination. One thing I might suggest is that you obtain a copy of some free software that allows you to calculate acoustic modes of rooms and other volumes. I have one of these, and if you’d like, I can upload it. I’m not sure how well this will work, but you could plug in different space dimensions as trial and error, and the program will calculate the frequencies of the acoustic modes. A serious issue, however, is the partially opened grille, which this program I don’t think will accommodate. Perhaps there are programs available that allow as input porous structures. One thing can be said right away is that all of these vibrational modes will have wavelengths that are of the order of the key dimensions in your space, and simply knowing the overall size of your concertina, you can calculate very easily by hand these wavelengths. You will probably conclude that all the amplified frequencies will be around and above several hundred Hertz. This is because solid walls give a zero velocity boundary condition, meaning that half wavelengths will fit between the key dimensions. For instance, for a 8-inch distance, or 8-inch half wavelength, a full wavelength will be 16 inches, corresponding to a frequency of 770 Hz. I’d think that an 8-inch length would be a maximum of the characteristic dimensions in question, and thus all other dimensions will lead to higher frequencies. It’s thus easy to see that any effect of this approach will be to influence the higher harmonics in most of the musical tones.

 

Clearly your warning "The small size of the instrument, however, may restrict how effective such devices would be for the lower notes, or may require designs that differ substantially from those found in accordions" was not misplaced. The full wavelength of the lowest note in a C/G is 264cm, about 8.6 ft, and a quarter wavelength of that might just fit in a large accordion.

 

I did the quick experiment because of late I have been puzzling over whether the overall box size might be significant in the strength of the lower partials. And this was brought on by my making a concertina with a significantly smaller square area than previously (with all the advantages in pump pressure of a small concertina) but with maximum dimensions that were not significantly reduced. This sounds impossible at first but if you consider how much smaller the square are of an octagon is compared to a hexagon with the same across the flats measurement all will be clear.

 

Before reading your post I was solely thinking of the width across the action box as being relevant as a resonator dimension, your post and maybe "ardie"'s as well caused me to think of the height of the action box as maybe needing consideration.

 

It is not often an experiment is so easily performed and in such a way as to leave one confident the results were not being confounded by some other factor.

I took the data you provided, loaded it into Audacity, analyzed each sound, generated the sound spectrum, exported it to a text file, entered it into a spread sheet, then massaged the data. The results are attached. You can look at this approach and probably manage to do it yourself when you want. In this spread sheet, you will see the fundamental frequency, as measured, with the name of the note. You should verify that these are correct. The Ordinate on the graphs give a normalized response such that the fundamental will have an amplitude in zero decibels. There are more than one graph more most of the pairs of sounds, with different graphs showing different Octave ranges.

 

My cursory analysis looking at these spectra is that there is indeed only little difference between the control and the modified instrument, and it’s not surprising that they sound the same. However, the graphs do show that there's somewhat of a consistent trend in that the higher harmonics are a tiny bit more prevalent with the second sound of each pair, exceptions being for the higher notes, where there seems to be a toss up. If I had to guess, I’d say that you placed the modified instrument sounds first, with the control sounds second. Please let me know if this is true, because it would then indicate that this Fourier Transform method of analysis may be more sensitive than my (and your) hearing. Take a look at the spectra yourself and see what you think.

Quite right, it was the first! Does this export data easily load into an xls or does it need to be loaded by hand?

If you’d care to take the time, for homework , you can pick off on these plots the harmonics that show significant difference between the two sounds, writing down the frequencies. You might see certain frequencies repeated. If so, you will have an idea of what resonances, or reverberations are occurring in your instrument; i.e., a “fingerprint,” if you will. Then, you can convert these frequencies into a wavelength, then to a half wavelength. You can then make various measurements inside your space and see if you can identify those half wavelength measurements with key dimensions in your space. If all is successful, you might get an intuitive feel for what particular dimensions are more important than others. Good luck.

One of my instrument designs does have a particular note that is always louder, perhaps this approach would help me identify the reason and lead to a better solution than making the reed less efficient and putting up with the resulting tone change.

 

Thanks for your thoughts and send that xls along..!

 

Chris

Edited March 12, 2011 by Chris Ghent