Acoustical effect of back woods

Talk about musical instrument construction, setup and repair.

Moderators: kiwigeo, Jeremy D

Post Reply
User avatar
Trevor Gore
Blackwood
Posts: 1426
Joined: Mon Jun 20, 2011 8:11 pm

Acoustical effect of back woods

Post by Trevor Gore » Fri Apr 05, 2019 7:57 am

According to the Journal of the Acoustical Society of America (of which esteemed body I am a member) one of the most downloaded articles from the last year published in The Journal of the Acoustical Society of America is this one:

“Effect of back wood choice on the perceived quality of steel-string acoustic guitars” (Samuele Carcagno, RogerBucknall, JimWoodhouse, Claudia Fritz and Christopher J.Plack). Some fairly famous names there.

https://asa.scitation.org/doi/pdf/10.11 ... ?class=pdf

This article has been made free to read, download, and share for a limited time (apparently).

After 16 pages of intense analysis their conclusion is:

“The results of our study indicate that steel-string acoustic guitars with backs and sides built using traditionally prized, expensive, and rare woods are not rated substantially higher by guitarists than guitars with backs and sides built using cheaper and more readily available woods. The poor ability of guitarists to discriminate under blinded conditions between guitars with backs and sides made of different woods suggest that back wood has only a marginal impact on the sound of an acoustic guitar”.

User avatar
lamanoditrento
Blackwood
Posts: 283
Joined: Thu May 05, 2016 9:50 am

Re: Acoustical effect of back woods

Post by lamanoditrento » Fri Apr 05, 2019 9:27 am

Heresy! Burn 'em at the stake :P
Trent

User avatar
peter.coombe
Blackwood
Posts: 589
Joined: Fri Oct 15, 2010 2:52 pm
Location: Bega, NSW
Contact:

Re: Acoustical effect of back woods

Post by peter.coombe » Fri Apr 05, 2019 10:27 am

Thanks Trevor. This is the first time I have seen the original article, and a quick read of it has not changed what I have said about the summaries of the paper elsewhere. I still think their methodology is fundamentally flawed. Using a rating system for 6 guitars that sound more or less similar and with no reference guitar is bound to give a randomised result, i.e. the methodology is designed to give the result they got. The participants are not given a fighting chance of any success. The reason for that is human auditory memory is very short so the only really reliable method of comparing music instruments is a one on one comparison with as short an interval between instruments as possible. That is how we all should rate our instruments, preferably one instrument being a known reference. If you think you can remember sounds for long periods of time you are kidding yourself. It is very easy to be fooled, particularly without a known reference. If the methodology is flawed, then their conclusions are a load of bull. It is easy to design double blind tests that eliminate bias, but also design it to fail.
Peter Coombe - mandolin, mandola and guitar maker
http://www.petercoombe.com

blackalex1952
Blackwood
Posts: 627
Joined: Tue Aug 27, 2013 6:36 pm
Location: North East Victoria

Re: Acoustical effect of back woods

Post by blackalex1952 » Fri Apr 05, 2019 5:43 pm

I have been playing and assessing acoustic guitars for many years now. Has anyone noticed a "compressed" sound from blackwood B&S guitars? It seems to me to be a bit similar to using an audio compressor...
Also as a Gypsy Jazzer, there seems to be a correlation between what is referred by players as a dry sound with the laminated B&S GJ guitars (think reverb) and a wetter sound with the solid B&S models.
I have also noticed a correlation between maple B&S guitars taking longer to "open up" and are generally less bassy. So any comments? cheers Ross
"Everything I say on the topic is based solely upon inexperience and assumption!"

User avatar
Mark McLean
Blackwood
Posts: 725
Joined: Thu Apr 10, 2008 2:03 pm
Location: Sydney

Re: Acoustical effect of back woods

Post by Mark McLean » Fri Apr 05, 2019 7:50 pm

I was really interested to see a report about the findings of this research paper a while back. Thanks Trevor for the link to the full manuscript. I am fascinated by this sort of approach to our craft. I also bring to it the viewpoint of a scientist - my day job is a medical academic, and I have been an editor of scientific journals for more than a decade. Even in my supposedly highly scientific discipline I often see people responding to new evidence with an attitude which could be summarized as "this study is obviously crap because it shows something that I don't believe". And fair enough - some studies are crap. So we need to evaluate the evidence.

In the field of luthiery we have a history of being fairly non-evidence based. Everybody just knows that instruments built with Brazilian rosewood are totally AWESOME, while sapele and walnut are OK but a bit ordinary. So, there is a lot of merit in this study just for bringing that idea to a careful evaluation. They have set up a good experiment by crafting 6 instruments which are otherwise identical, but with backs and sides of different timbers (Brazilian rosewood (BRW), Indian rosewood (IRW), maple, sapele, mahogany and walnut). The evaluation is done by 52 experienced guitarists (but you could criticize the fact that 51 of 52 were male). They played the instruments in carefully blinded conditions, even paying attention to blinding minor cues to timber identity such as smell.

There were two aspects to the reviewing. The first was a comprehensive evaluation of the sound of each guitar based on 16 criteria (e.g. brightness, volume, sustain, balance, "headroom", mellowness - each rated on a 1-5 scale). Bottom line was that the total ratings across the group did not identify any timber as being better (or worse). Now , I do get Peter's criticism of this methodology. Taking overall rankings of multiple aspects and averaging them out might bias the analysis towards a negative result - by averaging opposites. However, the fact that there was no "best guitar" is a interesting finding.

Peter suggests that a head-to-head comparison or a "reference guitar" approach would be a more valid method. Especially if they were compared in quick sequence, to get around the dodgy performance of human memory for qualitative aspects of sound. Well - in a second set of experiments the authors of this paper did actually do something like that. This was the "ABX experiment". They got 31 guitarists to play two guitars (A and B) in quick succession - about 1 minute each, with quick turnaround (using all of the same blinding methods). They then gave them one of those guitars back again (guitar X) and asked them to identify whether this was guitar A or B that they were playing again. No questions about quality - it was just a question of whether they could even tell the difference between two instruments. In this experiment only BRW, sapele and walnut were compared, and each player did it 4 times. Bottom line - there was very poor ability to tell which guitar they were playing, and generally no better than a coin-toss for accuracy.

Is this a perfect study? No, you can always pick some holes. But it is a very creditable effort to scientifically address an important question to our core business. I think it is very believable. In fact, probably correct. I am pretty sure there is a lot of bunkum in the "conventional wisdom" about different tone woods (OK - I confess that was my own a priori bias).

I am now heading up to my shed to continue making my latest guitar which has back and sides made of high-pressure laminate (that is basically laminex - the same stuff that your kitchen benchtops might be made of). I am very confident that the resulting guitar will be indistinguishable from BRW (except when you sand it and sniff it - the stuff smells like shit).

johnparchem
Blackwood
Posts: 414
Joined: Mon Jan 21, 2013 2:59 am
Location: Seattle
Contact:

Re: Acoustical effect of back woods

Post by johnparchem » Sat Apr 06, 2019 1:17 am

....The poor ability of guitarists to discriminate under blinded conditions between guitars with backs and sides made of different woods suggest that back wood has only a marginal impact on the sound of an acoustic guitar.
I question the study not so much for the scientific methodology and am happy with the details they provide. But the conclusion is too broad. I think it should have been limited to not being able to show "that the back wood choice does makes a significant and immediately obvious difference to the sound of a guitar. "

It looks like a good builder with a good set of tonewood setting out to make guitars that sound the same and using a steel string guitar design that does not have an active back was able to do just that build guitars with the same qualities. So the study did show that there were no magical qualities in the different tonewoods and sonically comparable guitars can be made using a design that worked with the physical characteristics of all the tonewoods in the group.

I am assuming a non active back design because when I looked at the Bridge admittance and modal parameters I could not even see the contribution of the back in Figure 1B They labeled the third strong peak as the transverse bending mode (cross dipole?). I could be wrong in my assumption that the backs resonant frequency should show up in their test of velocity response with a laser-Doppler vibrometer through coupling of the top and the back.

I used all of those tonewoods except maple and if I am able to modal tune the guitars I can not hear a distinguishable difference. I do hear a difference between the guitars I build with an effective active back and those with a reflective back. I do run into tonewoods like African Blackwood that are too heavy to make an active back. In that regard the selection of the backwood does have an impact.

User avatar
peter.coombe
Blackwood
Posts: 589
Joined: Fri Oct 15, 2010 2:52 pm
Location: Bega, NSW
Contact:

Re: Acoustical effect of back woods

Post by peter.coombe » Sat Apr 06, 2019 9:24 am

Interesting observation. If the guitars did not have active backs then maybe there is little or no difference in sound, as you would expect. However, confining my criticisms to the rating test, which is the main part of the paper, I still believe it to be fundamentally flawed. They get a negative result, i.e. the results are randomized. That could be due to a number of reasons
(1) The participants could not tell the difference because there are no discernible differences
(2) The methodology is flawed
(3) People have different tastes in sound

The authors go for (1). However, there is nothing in the paper that eliminates (2) & (3), and - there is no control. A control is an essential part of any scientific test, really really basic stuff. They spent a lot of effort to eliminate any bias, and I have no problem with that, but no control, come on!! We are supposed to assume that the test as designed is valid and does actually work. With no control test, there is no indication that is actually the case, so it is a big assumption we are supposed to make with no evidence. Both (2) and (3) can give you the same result and with no control you cannot eliminate either one of them. In fact given the limitations of human hearing memory, (2) is quite likely to be true, and we all know (3) is true. One example of (3) is I had two guitars a year ago that sounded very different, one Rosewood, the other Sassafras. Blindfolded anyone would be able to tell the difference easily. However, I had real trouble rating the best guitar, and potential customers were around 50/50. Put those two guitars and my potential customers into their test and they would have concluded there was no difference. That is bullshit. The ABX experiment is more convincing, at least the participants have a fighting chance of getting it right.

Call me pedantic, but I have peer reviewed scientific papers in the past (I have a PhD) and this sort of thing comes up all the time. No control -> evidence is questionable at best -> no conclusion possible. They may as well throw the rating test in the bin, despite all the effort they put into it.
Peter Coombe - mandolin, mandola and guitar maker
http://www.petercoombe.com

User avatar
Trevor Gore
Blackwood
Posts: 1426
Joined: Mon Jun 20, 2011 8:11 pm

Re: Acoustical effect of back woods

Post by Trevor Gore » Mon Apr 08, 2019 10:25 pm

One can get very picky about the way these sorts of experiments are performed, but I'll try to stay away from that. However, trying to identify differences in sound due to the back woods when the T(1,1)1's and T(1,1)2's range over a semi-tone on non-live back instruments (no T(1,1)3's apparent, Fig 1A) doesn't give you much of a chance.

How the back and sides are treated can make a huge difference to the way a guitar sounds. Live vs. non-live backs, mass loaded sides etc. are ample proof of that. But what wood you use, much less so, UNLESS you systematically rely on the average species properties rather than build technique to effect the differences.

What you hear when a guitar is played are the modal resonances, defined by their center frequencies, amplitudes and bandwidth. If you can measure the mechanical properties of the wood (density, Young's modulus, damping) and use the right, well documented techniques, you can match the important modal resonances of guitars made of different woods pretty closely. That's what builders are doing (whether they realize it or not) when "The guitar maker, by treating each back in the way that his experience suggested was best, has to a very large extent compensated for any physical differences between the types of wood". So a wide range of woods can be made to sound alike, which means, of course, that players can't pick the difference, and guitar builders have a much greater range of woods at their disposal. I know that I can't pick wood species on my (modally tuned) guitars. All we need to do now is stop the factories from perpetuating the species myths and get them to promote some of the more sustainable alternatives. :D

Dave M
Blackwood
Posts: 287
Joined: Tue Jul 15, 2014 6:44 am
Location: Somerset UK

Re: Acoustical effect of back woods

Post by Dave M » Tue Apr 09, 2019 2:18 am

Trevor. Quite.

Dave
------------------
Dave

User avatar
nkforster
Blackwood
Posts: 108
Joined: Tue Oct 30, 2012 1:02 am
Location: Brisbane
Contact:

Re: Acoustical effect of back woods

Post by nkforster » Tue Apr 09, 2019 8:58 am

A more useful study might be to see how easy it is to sell a guitar for decent money if it isn't made from traditional woods. And then to see how easy it is to sell 20 or more guitars a year made with non traditional woods for decent money. Spoiler alert: It's not easy.

This experiment, flawed or otherwise is an attempt to encourage guitar buyers to consider buying a guitar made from non traditional woods. If we get lost in how thorough the methodology is, we miss the point of the study. The point of the study is to help steer buyers away from traditional exotics. There was a similar experiment recently in the violin world, comparing the tone of modern violins to "old master" violins. Yes, the methodology was suspect, but this is often the case . It doesn't matter unless you think the purpose of the experiment is to reach some sort of unassailable truth.

These experiments are about business. Not sound. Or tuth.

How successful they are is at converting people into customers s another matter.

Just my two-penneth worth...

User avatar
peter.coombe
Blackwood
Posts: 589
Joined: Fri Oct 15, 2010 2:52 pm
Location: Bega, NSW
Contact:

Re: Acoustical effect of back woods

Post by peter.coombe » Tue Apr 09, 2019 9:46 am

Yep, Nigel you are right. Rosewood guitars sell, other woods are harder to sell. Hopefully this study and the publicity it has received might start to shift that, but it is like shifting the Titanic with a tinny. I have not had the same problem with mandolins. Australian natives sell just as well as the traditional Spruce/Maple.
Peter Coombe - mandolin, mandola and guitar maker
http://www.petercoombe.com

User avatar
nkforster
Blackwood
Posts: 108
Joined: Tue Oct 30, 2012 1:02 am
Location: Brisbane
Contact:

Re: Acoustical effect of back woods

Post by nkforster » Tue Apr 09, 2019 10:43 am

peter.coombe wrote:
Tue Apr 09, 2019 9:46 am
Yep, Nigel you are right. Rosewood guitars sell, other woods are harder to sell. Hopefully this study and the publicity it has received might start to shift that, but it is like shifting the Titanic with a tinny. I have not had the same problem with mandolins. Australian natives sell just as well as the traditional Spruce/Maple.
We'll have to agree to agree Peter! I think that these experiments are also about securing funding to build guitars. It was this or another similar experiment a few years ago where EU funding was secured to build a few guitars to see if domestic woods made decent guitars. Fair play. If you can get the government to cough up to subsidise my living, count me in!

Things like CITES help the public taste change. Once big makers stop using rosewoods (because the paperwork is such a PITA) they'll soon come up with reasons why domestic woods are just as good. A couple of years back, Bob Taylor made a great video explaining why its not reasonable to keep insisting that all ebony should be jet black. The same sort of thinking and marketing will apply to using non exotics fairly soon. Once the big players get behind it, it'll make life easier for the one man shops like ours.

And yes, finding niche markets that aren't so set in their ways is one way around the issue. But as you know, if you were aiming at the very top of the Bluegrass market (which is very lucrative) the materials you could use would be decided by the market, not my the maker. In the Celtic mandolin market where I operate, there is less money, and the market is more open to ideas. Just as long as they are cheap!

hepikohetaniwha
Beefwood
Posts: 26
Joined: Sun Oct 14, 2012 7:52 am

Re: Acoustical effect of back woods

Post by hepikohetaniwha » Tue Apr 09, 2019 7:01 pm

It’s an interesting topic, and I’m curious Peter; what would be the control that the study design omitted? A control group, in my high-school level understanding of science, is the experiment run the exact same way but without the variable that is being tested. Here the variable is back wood species. How would we design an experiment in which that variable was eliminated?

User avatar
kiwigeo
Admin
Posts: 9656
Joined: Sat Sep 29, 2007 5:57 pm
Location: Adelaide, Sth Australia

Re: Acoustical effect of back woods

Post by kiwigeo » Tue Apr 09, 2019 9:32 pm

I was going to ask the same question.

Perhaps a guitar with no back would be the control?? :)
hepikohetaniwha wrote:
Tue Apr 09, 2019 7:01 pm
It’s an interesting topic, and I’m curious Peter; what would be the control that the study design omitted? A control group, in my high-school level understanding of science, is the experiment run the exact same way but without the variable that is being tested. Here the variable is back wood species. How would we design an experiment in which that variable was eliminated?
Martin

User avatar
peter.coombe
Blackwood
Posts: 589
Joined: Fri Oct 15, 2010 2:52 pm
Location: Bega, NSW
Contact:

Re: Acoustical effect of back woods

Post by peter.coombe » Wed Apr 10, 2019 10:15 am

Scientific hat on.

When doing scientific experiments there are two questions you need to ask and answer
(1) Is what has been observed in the experiment real and not an artefact of the methodology
(2) Is it repeatable.

A control is used to answer (1), and you repeat the experiment a number of times and use statistics to answer (2). Statistics will give you a probability of the observation happened by chance. The more times you repeat the experiment the better since that will give you more degrees of confidence it is not a random event. Usually a probability of less than 5% is accepted as being significant, although 1% or less is better.

The study we are discussing has (2) reasonably well covered, but (1) has not been covered at all. Now the guitars in question were all dead back and with similar measurements, so we would expect they would all be difficult to tell apart. If we added a live back guitar with different measurements and randomised it with the other guitars then that would be our control guitar. We would all expect it would sound different and would be easy to tell apart. If the participants gave that guitar a different score then (1) would be answered. If the score was randomized with the other guitars then the methodology is not working and the conclusions are invalid. The result is an artefact of the methodology. If the methodology is working properly I would expect that control guitar to stick out like a sore thumb. What the control guitar does is to demonstrate that the participants are actually able to pick differences using the methodology. That was missing in the study so you can't have any confidence in the methodology at all.

When you get a negative result in an experiment your control needs to show a positive result to prove that the negative result is not because the methodology is not working. With a positive control, the methodology is then shown it is actually working. Conversely if the result is positive, you need a negative control.

Nigel, my first reaction to the study was - lucky bastard, he managed to get a research grant to make 6 guitars. Woo Hoo!
Peter Coombe - mandolin, mandola and guitar maker
http://www.petercoombe.com

hepikohetaniwha
Beefwood
Posts: 26
Joined: Sun Oct 14, 2012 7:52 am

Re: Acoustical effect of back woods

Post by hepikohetaniwha » Wed Apr 10, 2019 10:47 am

I think I follow...

So it’s a bit like saying, in order to see if you can tell the difference between Granny Smiths and red delicious (blindfolded perhaps), then first let’s make sure you can tell the difference between apples and bananas?

jeffhigh
Blackwood
Posts: 1499
Joined: Thu Feb 07, 2008 5:50 am
Location: Caves Beach, NSW
Contact:

Re: Acoustical effect of back woods

Post by jeffhigh » Wed Apr 10, 2019 6:00 pm

I agree with a lot of Peter Coombe's concerns about the methodology of this study.
In addition
-what the guitarist hears of the guitar is dependant on room acoustics
- The fact that the guitarist hears the instrument as he is playing it means that they can make subtle adjustments to playing style to get "their sound" and minimize variation. If they were wearing effective earplugs and later heard a recording taken from an audience perspective it may be more meaningful and would allow for rapid change between instruments for real time comparison.

User avatar
peter.coombe
Blackwood
Posts: 589
Joined: Fri Oct 15, 2010 2:52 pm
Location: Bega, NSW
Contact:

Re: Acoustical effect of back woods

Post by peter.coombe » Wed Apr 10, 2019 8:50 pm

So it’s a bit like saying, in order to see if you can tell the difference between Granny Smiths and red delicious (blindfolded perhaps), then first let’s make sure you can tell the difference between apples and bananas?
Yep, something like that.

Negative results in science can be tricky. If you repeat the experiment enough times you might get a positive result. Sometimes that is very significant and changes everything, other times it is just a point on the tail of the normal distribution. In this case the tricky bit is how different does the control guitar need to be? If it is a mandolin, then that does not prove anything. If it is the same as the guitars in the test then that won't prove anything either. Somewhere in between is what you want, but where about in between? Ideally you would have a series of control guitars covering in between and the point at which the participants can tell them apart then that gives you a measure of the sensitivity of the methodology. That can be very useful information not just for this study, but also for others that might follow.

These sorts of studies are difficult because there are always limited resources in terms of the instruments available and the number and quality of the participants, and time available, so compromises have to be made. It is a matter of judgement as to whether the compromises are big enough to invalidate the study, and there will certainly be disagreements about that. Using a control guitar is not so difficult and it is a great pity they did not do it. It would have made their case much more convincing.

Another, perhaps more interesting study would be to use live back guitars.
Peter Coombe - mandolin, mandola and guitar maker
http://www.petercoombe.com

johnparchem
Blackwood
Posts: 414
Joined: Mon Jan 21, 2013 2:59 am
Location: Seattle
Contact:

Re: Acoustical effect of back woods

Post by johnparchem » Thu Apr 11, 2019 11:38 pm

Thanks Peter, On your first post I had trouble imagining what the control would be. I appreciate your detailed explanations. My understanding from your comments is that without the control, testing with instruments that clearly sound different, we have to assume or be convinced that their methodology would have given them positive results with different sounding instruments because they do not demonstrate it.

User avatar
peter.coombe
Blackwood
Posts: 589
Joined: Fri Oct 15, 2010 2:52 pm
Location: Bega, NSW
Contact:

Re: Acoustical effect of back woods

Post by peter.coombe » Fri Apr 12, 2019 9:48 am

Thanks Peter, On your first post I had trouble imagining what the control would be. I appreciate your detailed explanations. My understanding from your comments is that without the control, testing with instruments that clearly sound different, we have to assume or be convinced that their methodology would have given them positive results with different sounding instruments because they do not demonstrate it.
Correct. It is an assumption because without a control it is difficult to be convinced the methodology does actually work. Standard practice is to calibrate the methodology first and that becomes the control experiment, then there is no question about the methodology, you know it works and how it works and how sensitive it is. However, sometimes calibration is difficult or impossible to do and in this case it would be very difficult, and impossible if you don't have the resources (the guitars).
Peter Coombe - mandolin, mandola and guitar maker
http://www.petercoombe.com

johnparchem
Blackwood
Posts: 414
Joined: Mon Jan 21, 2013 2:59 am
Location: Seattle
Contact:

Re: Acoustical effect of back woods

Post by johnparchem » Tue Apr 16, 2019 9:52 am

I held off giving my own from my hind side comments on blind listening tests mainly because of where I am pulling them from. Here goes ... Part of the context for my point is that humans are not like a microphone and a computer, we do not hear reality rather we have learned through evolution and in our childhood, to process a relatively slow low bandwidth organic connection between our ears and our brain pretty high fidelity sound. It seems like part of that ability is that we construct the high fidelity sound in our brains from the limited information we get. (this is part of my limited understanding after a drink enabled conversation with a neuroscientist studying that interface.)

Also context - What we can hear as an adult is a learned skill we mostly picked up in our childhood. As we grow the body prunes many neurons and we lose some of our innate abilities to hear. I ran into this learning Thai as an adult. What sounds like the same word to me can be 4 to six different Thai words. For the native speaker those individual words sounds as different as night and day. So much so that as I learned Thai a small mispronunciation for example a mid sound vs a high sound on the same english vowel was hard for a native speaker to understand even in context unless they were exposed to what english speakers do to their language

Going to the first point in terms of distinguishing an instrument we probably first hear Guitar - then Classical, electric and steel string acoustic - then in each of those groups people probably hear the player way more than the guitars - next might come bracing patterns (i.e. x brace vs ladder) and so on. I think the farther one goes down the tree I laid out away from guitar the ability to get a positive result in the blind test goes down. We do not know as per Peters critique of the study was that they had no control to test for a positive outcome. The researchers did no work to distinguish what level of difference would provide a positive outcome for the test.

There may be tests with positive outcomes. Maybe a tune played on one guitar vs another would get a longer average listening time if posted on youtube or itunes. Remember the switch to CDs or MP3 it might have also been hard in a blind test to tell the difference yet there was a different in how people felt about the music over the long term.

All in all I am with Peter Coombe without a control it is hard to tell what this test shows. The best they can say is the test did not demonstrate that there is a disquirnishable difference.

Post Reply

Who is online

Users browsing this forum: No registered users and 14 guests