Testing John Schneider's tailpiece design
Edit: Experiment stopped. There were not enough responses. I failed to reject that there is no significant difference between the tailpieces.
tl;dr: go here and listen to and/or download the recordings. Then go here and vote.
John Schneider designed a tailpiece described here (and here). He made one and sent it to me, at my request to make comparative recordings.
The recording environment was a bedroom with a mixture of hard and soft surfaces. The microphone was a single Røde NT1-A in approximately the middle of the room at neck height. I stood in the same place for all recordings, about 2 feet away facing the microphone with the microphone about 45 degrees left of the bridge.
Strings were Thomastik Dominant Medium G, D, and A. Warchal Amber Medium E. I usually use a Goldbrokat E, but I didn't have one with a ball end. No fine tuners.
First I installed John’s tailpiece as it was provided and put the copper foil in as instructed. Tailpiece distance from the bridge was within his specifications. I made a recording of it. Next, I took out the copper foil and made another recording. Then I installed my boxwood tailpiece (without my usual fine tuner) and made a third recording.
For each recording, I played a chromatic scale, aiming for 4 notes about 2 seconds each, down up down up, from G3 to E6. The first two notes I use for pitch correction and getting settled in a position. I only use the last upbow for analysis. I kept the bow the same tightness, and have not rosined it for days before. Still, it is difficult to maintain the same pressure and a certain playing distance from the bridge.
One wolf tone was found at C6 using the plain brass tube. Adding copper foil alleviates it somewhat. This was already present, although weak and limited to the higher partials, on the setup with my boxwood tailpiece. My usual setup with a Goldbrokat E and (short) fine tuner does not exhibit this wolf tone. See the graph below.
C6. From left to right: boxwood, plain brass tube, brass tube with copper foil. Notice the relatively constant amplitude of the fundamental with pulsing limited to the upper partials in the boxwood tailpiece, the pulsing of all partials of the plain brass tube, and the same but less severely on the brass tube with copper foil.
A note about analyzing recordings: It is difficult (for me) to make a consistent sound on a violin. Within a single bow stroke one can expect a 3dB variation in the amplitude of the fundamental, so any differences less than 3dB are insignificant.
The first thing I observed was RMS. There is no significant difference in the amount of sound that reaches the microphone across all three setups. The setup with copper foil averages -39.3 dB; no copper foil averages -40.6 dB; boxwood averages -40.1 dB. This was somewhat of a surprise to me as I honestly felt like my violin was louder with John's tailpiece. However, I am aware of the power of placebo and that energy could have been redistributed to partials to which I'm more sensitive.
Next, I looked at the graphs of G3, B3, D4, D#4, A4, C#5, E5, G#5, B5, and E6 in my spectrum analyzer (Voxengo SPAN). I avoided C6 as that had a wolf. I could not see any consistent significant differences. Sometimes John's tailpiece(s) would exhibit a stronger fundamental, other times stronger overtones, or weakening or strengthening of particular partials. There was nothing consistent I could make a conclusion on. Here is a sample of graphs. They're different colors because I color coded to stay organized.
A4, boxwood:
A4, plain brass tube:
A4, brass tube with copper foil:
Even though I can't generalize about overall sound of these tailpieces, I can hear a difference between them given the same note. Also, different people will find different things pleasing. Therefore, I have decided to share my recordings to Violinist.com members, with a poll asking which recording they like most. One is of my boxwood tailpiece, one of the plain brass tube, and one of the brass tube with copper foil. Only I know which recordings are of which tailpiece. I have cut off the recordings before C6, as the degree of wolf will identify the tailpiece.
To participate please go here and listen to and/or download the recordings. Then go here and vote.
I'll check back in a week or when I remember.
Edit: A was the brass tube with copper foil. B was the brass tube. C was boxwood.
Replies (29)
I made them .ogg q10, transparent to almost everybody. Now you can preview.
John Schneider,
The 4th partial in C6 in the boxwood isn't part of the wolf, because it isn't regular like in the 8th partial (between 8k and 9k). It might have been bad bow angle or bad contact because of other partials pulsing. Anyway, pulsing like that might sound louder (it actually sounds weaker to me), but it's just one note and not one I played when I thought my violin was louder.
In audio recording, sound pressure level is recorded in decibels relative to the maximum sound pressure level before clipping, defined as 0. In the spectrum analyzer graphs, higher is closer to 0, which is louder. I can't say anything is better or worse, which is why I started a poll.
I went back and forth among the recordings. A couple recordings sounded like a few of the notes had some "vibrato" in them. Which was the best sound? >shrug<
Pulsing that is spaced at regular intervals is usually two things resonating close to the same frequency and in phase. You see a lot more of that in the partials of the brass setups.
Try double stopping octaves and recording that. If you are not spot on, you should see a similar effect in the primary resonance at least.
It is not an unpleasant sound, IMO, but I think most violinist would prefer to add their own articulations, rather than have the violin decide it for them.
If the frequency line is broken up, that could mean a few things, including a weak wolf tone.
The brass setup shows a lot of breakage in the upper partials. But by then, the loudness of the partial is low enough that it might only show up as a change in timber to a careful listener rather than and an ugly overtone.
A bad wolf tone would show lots of breakage in the lower partials.
Thanks for the test. Very interesting.
I tried to keep my finger still. Maybe this "vibrato" was me adjusting the pitch or bow bounce. I tried to play a straight tone for at least the last upbow, and the last two notes most of the time.
One thing I forgot to note was that tuning is quite difficult on the brass tube, because of lack of a fine tuner on the E string and that the tension of one string affects all the others.
Dear Harvey: I thank you for your enlightening me about how to read the set of three decibel charts you published. Again, it was kind of you to go to all that effort for us. As I mentioned yesterday, I tried to find a way to evaluate the message your charts present. I came up with the idea that counting the predominance of peak positions within given ranges on said charts would decode what they were showing to us. When counting the peaks apparent in the graphs over the value of 81 I find that my tail piece with foil (that is the preferred and only assembly I recommend) has a 5% greater preponderance of peaks present when compared to the boxwood tail piece. As the charts progress upward toward the greater dB readings, at 66 on the vertical scales my tail piece with foil shows a 20% greater total when counting the apparent peaks above that value. The highest dB reading of those peaks which I inspected was over 60. My tail piece shows a 40% greater preponderance of peaks over 60 on the vertical scale. I believe this shows my tail piece gives a violin more power. As to the first chart where you show the pulsing of the sound being so much more pronounced in the violin's output using my tail piece - I now realize that this pulsing result indicates what a person hearing that note would recognize as resonance in that note. These results are in perfect keeping with what my auditory experience has been as I have tested this design over the years. My tail piece design provides violins with a more powerful and more resonant sound production. Best regards and again thank you for your charts which clarify what I have been saying have been my results. John
John,
I wouldn't be so quick to conclude that the number of partials over a certain SPL is indicative of power, or likability. And that is just one note. As I stated before, average power was the same across all tailpieces. The only valid analysis I can think of (in this situation where differences seem so small and inconsistent) is to note the levels of each partial for each note up to 20000Hz (where most people can't hear anything anymore) and do some comparative statistics, which I don't know how to do. Even then one might not find anything meaningful. Therefore, the quickest way is just to get humans to listen to the sound and decide on that. That's the proof of the pudding anyway.
John, let me talk about decibels.
Spectrographs typically use something called the dBfs system. Basically, 0 represents the loudest value (0= fULL sCALE) it can store, and everything else drops off along a negative scale.
It is a logarithmic scale that measures amplitude. For every drop off of 6db, the amplitude is halved. So a spectrogram with a dB scale can give a skewed perspective on how strong the partials are.
When you switch to power, the comparison gets worse as power is proportional to the square of the amplitude. So for every 6dB drop off, the power drops by 1/4.
As an example, going from a high of -50dB to a low of -80dB is a drop off of 30dB. That is an amplitude drop of 1/32 (97% decrease) and a power drop off of 1/1024 (99.9% decrease).
In terms of power, anything beyond the first few peaks would not contribute much at all to the loudness, but it might affect the timber (color or quality of the tone).
Dear friends: It is hard if not impossible to detect how powerful a violin sounds unless you are in the same room with it or you have a decibel meter. That is why I selected a way I thought could compare tail piece results from Harvey's three dB graphs. Apparently that won't work based on some of your comments. However, no one corrected me about my suggestion that the pulsing seen in Harvey's first graph is a visual representation of the resonance in the note charted. As I daily sit and play I marvel at the power and resonance my violins give me which is far above that which I received from my old tail piece. I had hoped Harvey's efforts would settle those issues but apparently there is no easy way to show how much better I find my tail piece to be. I am not throwing in the towel but as we go along in this project it becomes more and more likely that only playing a violin with my tail piece in place will let any one individual share in my elation over my discovery. Perhaps Harvey's trial by listening project will provide some concrete results! Best regards, John
Well, about that pulsing. Most people call it a wolf, and most people find it objectionable. I don't know if that's what you hear or not, unless you point out a note where you find "resonance" and the spectrogram looks like my C6.
"As I daily sit and play I marvel at the power and resonance my violins give me which is far above that which I received from my old tail piece. I had hoped Harvey's efforts would settle those issues but apparently there is no easy way to show how much better I find my tail piece to be."
____________________
I believe that you believe your tailpiece is an improvement, which has also been true for most of the people who invented the plethora of other "violin improvements", so you probably don't need to repeat it again.
It might be more productive to look into the legend of Narcissus. (wink)
Dear Harvey, David, Carmen and John plus all others reading this: First of all, Harvey, I can honestly say that the vast majority of my string's frequencies produce a noticeable resonance. Especially the open strings are resonant and any fingered note that exhibits a sympathetic resonance with those open strings. But even most frequencies that are not sympathetic in their resonance to open strings none the less resonate. That is one of the two really enjoyable things that keep me playing every day. So, the pulses we notice on your first chart, which are much more predominant in my tail piece recordings when compared to the boxwood tail piece results, can be explained as resonance identifiers and not wolf notes - unless the violin you used is prone to wolf notes and I know you would never have permitted that. My luthier friend in England, now that he has my tail piece on his violin, commented not only on how much improvement he noticed in the overall tonal production it gave but also he noted that his G string took on a whole new personality by giving resonances he had never heard before from a violin's G string. I have the same observation to make about my two violins. I can say however that the Del Gesu copy enjoys less of that G string boost than my Strad copy. But both exhibit G string sounds so resonant that you could honestly liken them to the G string sound that a cello makes. I am surprised that none of the four of you who have received my gift have seemed to notice that G string improvement - or are you just too shy to mention it in polite company? (wink). Thanks again for all your efforts Harvey. Without these charts we would be at a dead end in our discussion. My playing enjoyment is also enhanced not only by the superior resonance I hear on all my strings but by the added power my tail piece gives to both my violins. I therefore wondered how that added benefit didn't seem to be obvious from a reading of your graphs, Harvey. As you recall yesterday I wrote that I had read the graphs you provided and studied just the peak dB data points. I noted that in tracking all frequency peaks over the dB values of 81, 66 and 60 my tail piece gave greater peak performance by as much as 40% in comparison to the boxwood tail piece. You folks replied that my evaluation couldn't be used to conclude that my tail piece produced more sound than the boxwood tail piece. So today I have spent an hour and done some more analysis on your dB charts you so kindly provided for us. To produce a completely creditable study I noted the dB readings from the vertical scale for each frequency data point plotted by your tester. In the case of the boxwood tail piece there were 45 data points in all and for my tail piece (as you say "with foil") there were 42 data points charted. I added all those values for each of the two tail pieces and came up with a total raw score of 2,841.5 for my tail piece and 3,113.3 for the boxwood tail piece. Since the lower score shows the most power present in the frequencies plotted you can say my tail piece proved to have produced more sound! If you average those total values over total data points the averages are 69.2 for the boxwood and 67.7 for my tail piece. Again, my tail piece provides the lowest average score so again it has produced the most powerful sound values. So, using either evaluation, my tail piece produced the most powerful frequencies during both the peak performances and also provided more powerful frequencies overall. I believe your graphs have provided us with the clarification we needed to show both the improved resonance and also the improved power my tail piece gives when compared to a standard tail piece. Best regards, John
Here is A4, an open string. Left to right: boxwood, brass with no copper, brass with copper.
The lower partials don't pulse on any of them. The 7th partial on boxwood might do it slowly, but not fast enough to be obvious to humans.
Carmen's explanation of decibels is valid. Counting peaks is almost meaningless. Here is average SPL for A4:
Boxwood: -40.7dB(FS)
Brass: -43.5dB
Brass with copper: -40.7dB
No difference between boxwood and brass with copper. The drop in SPL on brass was almost significant, but then again it's just one note.
Average power across all notes was the same on all tailpieces.
Harvey, what is the time resolution on your sonograms? It might be a knob setting with something like 1 sample/second to 100 samples/second.
Voxengo SPAN was set to average over 1 second. Block sizes are 2048 samples, so in your terms 21.53 samples per second. Spectrogram screenshots show about 1 second of sound. Not sure where time resolution is shown. Using Adobe Audition CS6.
Dear folks: I am wondering what good Harvey's dB graphs are if - as you all have been telling me - they don't mean anything? I noted that the peaks denoting dB measurements values for my tail piece went much higher on the graphs than the peaks for the boxwood tail piece. I was abruptly told that it didn't mean anything and could not be used to show my tail piece was producing more power in its notes. I then measured the dB ratings for each visible frequency shown and added them and averaged them to show my tail piece was overall producing more sound. I was again abruptly told that that evaluation was "useless". So, what is the purpose of Harvey even having made those dB graphs? What do they show and how can they be read? I have a decent education in statistics and have taught engineering courses at the University of New York for 11 years so I know how to read charts and graphs. The points on Harvey's graphs that evolve into vertical lines denote frequencies and the vertical scale gives the dB rating for each frequency point. Adding all those readings together gives a total dB value for all the frequencies plotted. Say what you will, there can be no other interpretation given to those graphs. They show my tail piece performing with more power than the boxwood tail piece. John
Oh dear! Oh dear! Oh dear!
If you really taught engineering at the University of New York (though a quick search doesn't support it), I feel really sorry for the students!
John, we are trying to be helpful but it doesn't help if you misrepresent what we post.
I explained the meaning of the dB scale in those charts and how they translate into sound power. You might want to reread that.
If the spectrographs had been displayed with a linear vertical amplitude scale instead of dBs (which is actually a logarithmic measurement), you would see how dramatically the amplitudes are rolling off.
Most of the sound is going to come from harmonics up to about the 12th partial (5,280Hz).
With that in mind, let's compare boxwood with brass tube+foil.
The main difference between the two lies in the third and fourth partials. The boxwood tailpiece has a much stronger 3rd partial than the tube. The tube has a stronger 4th partial than the boxwood.
Everything else is about the same between the two, which supports the observations of myself and Harvey that it is tough to notice a difference in timber between the two.
Turning to the sonograms (the brightly colored horizontal lines), what we are looking for here are potential trouble spots. Each partial (horizontal line) should have a consistent color as you go forward in time with very little coloring in-between the lines.
The boxwood shows a very strong, consistent pattern in the first few partials with some breakage in the 2nd and 4th.
The tube shows a marked, periodic pulsing in just about all the partials. I noted that I heard "vibrato" on a couple of the sound samples, although I should probably have used the term "tremolo", since it was more a loudness effect than a pitch wavering effect.
Because Harvey did not say which sound sample went with which tailpiece, there is no way I can confirm that what I heard explains what the sonograms are showing us.
My educated conclusion would be that there is no difference in the timber or power of the tone between the boxwood and brass tailpieces.
But in the specific case of the A4 note, the brass tailpiece is generating a stronger resonance with some multiple of 440Hz which is causing the pulsing seen in the sonograms.
Extend this conclusion to other notes and the violin in general at your own risk. Let's just say that some resonances of the brass tailpiece might be stronger than the boxwood.
If the violin is already strong at these resonances, then what might be pleasing to your ear would probably be considered annoying to an accomplished violinist who wants full control over the articulation of the notes.
If the violin is weak in these resonances, then it might possibly give an acceptable boost to these tones.
Thank you Carmen and John for your patient and kind responses. To clarify my teaching credentials for some skeptics I can say I am not on a list of active professors at SUNY. My tenure was back in the 1980s. I hope that clears up the matter for those few who are looking for ways to question my veracity. Best regards, John
>> And that is why Stradivari chose not to use electronics
I have it on good word that he purchased a spectrometer, but gave up on it when he realized he had no electrical outlets in his workshop.
How did you come to the conclusion that the violin did not develop with/from theories?
Dear friends: I have visited many museums including the Smithsonian and Ashmolean ( Hills Brothers collection) and etc.. I viewed their displays of musical instruments including stringed instruments. I can correctly say that the museum experts who assembled these historic collections without exception have provided captions with all the displays including the stringed instruments. These captions explain to the public how the modern violins, etc. evolved from earlier forms popular in the middle ages and even before that. These museum curators are very learned in their field and all say modern forms are the result of evolution and great intuitive thinking, not theories born from mathematical equations. In those days few if any luthiers even knew how to use math to design improvements into their creations. We still don't!!! So, they all invented their own shapes or copied them from some other maker's success. They were all “cut and try” designs without any "high fluting" theories the likes of which I have been exposed to in copious amounts here on violinist.com. So John, you are quite right when you say violin design is not the result of theories. It is the result of trial and error - just like my new tail piece. I have tried for over 6 years to create some way of increasing a violin’s volume of sound and also its resonance. I have succeeded as my decibel meter tests and my 80 year old ears attest. Before someone protests and suggests I must be wearing a hearing aid, I do not. My hearing is perfect for any aged individual. I even hear clocks ticking on the wall when others much younger cannot. I am old enough to have been listening to violin playing longer than most of you have been alive. Much of it has been my own performances beginning at the age of 6. Even if the whole world of luthiers agree that I have failed I will listen to the successful testimony of my violin’s as I play them and I will know better. That just gave me a thought! How many luthiers can even play a violin well enough to be convincing? All that I have met can only scratch out enough sound to be sure they have put the strings on in their proper order and then tune their work pieces. Best regards, John
John S, museum curators are generally not the people who are consulted when one is seeking high-level violin expertise. The highest level of expertise pretty much resides in the fiddle trade, where it is a full-time pursuit.
We are no strangers to museums. For instance, John Montgomery is the professional violin tradesperson/consultant associated with the collection at the Library of Congress. Bruce Carlson is the tradesperson-curator for the Paganini Guarneri in the museum in Genoa. I was one of several consultant/tradespeople associated with the violin collection at the Henry Ford Museum, and one of my violins is in the collection of the Smithsonian Institution. So you're really barking up the wrong tree for violin expertise. Instead, ask the people who the museums ask.
Prevailing opinion these days seems to be that the Cremonese makers made use of math and proportion-based drafting systems which were popular at the time. You can look up concepts like "golden ratio", "golden spiral" etc. if you wish. There's also a rather respected book on the the mathematical and geometric principles behind the design of Cremonese instruments, written by Francois Denis.
From a description of the book:
"In this richly illustrated 256-page book, the author explains how the proportions of musical instruments are connected to an archaic definition of measurement. In the same way as Renaissance architecture, the dimensions of forms are related to the dimensions and intervals of our musical composition rather than to the more familiar divisions of our graduated rulers."
As to whether violin makers can play, I had to make a decision between the Curtis Institute, the Mittenwald Violin making school, and the Weisshaar shop at the age of 18. With that background, I'll assert that even if many violin makers don't play at paid performance quality, most of the better ones can at least torture a violin to a level that they can find out what they need to know. ;-)
John Cadd wrote:
"About theories--The development was gradual and slow. There was no sudden blueprint written out. It was much slower than the invention of motor vehicles."
____________________________
Your tea leaves or ouija board told you that?
I'm getting kind of bored of this. I only got 3 legitimate responses, and they weren't all the same, and that's not enough to conclude anything even if they were.
A was brass with copper. B was plain brass. C was boxwood.
I should have realized this before but if you claim to have improved the violin in some way, you should do this kind of experimentation yourself. "I believe it so it must be true" is meaningless. Placebo is powerful. Not even those perfectly aware of it can escape its effects.
I will edit the OP.
Dear Harvey,John & John,Carmen,David and all: Thank you for all your efforts on behalf of our discussion. It has been a good ride and I have learned a lot from everybody. I will take away many new realizations about the violin and about all you top notch luthiers. It is easy to see that you have studied in far more depth than I have. Thanks for sharing all that knowledge. I had no idea violins were so sensitive to each and every one of their component parts. I suspect even the glue has its own effect on the sound! Well, I will just keep enjoying my tail piece invention and will hold found memories of our several months together. Again, many good wishes to all of you. John Schneider
Say, did anyone ever think that it would be a good idea to mix Diatom powders in with your glue?? Just a final shot! John
Sure, but it doesn't seem to offer any advantage, and the abrasive qualities dull sharp-edged cutting tools rather badly.
The formalization of the scientific method and discovery through rational thought blossomed in the 17th Century to the point that it is called the period of the Scientific Revolution.
It is the time of Galileo and Newton.
The mechanical wonder we call the pocket watch was perfected.
I can go on about the marvelous inventions of the period, all of which required at least a usable theory of how things worked if not a perfect comprehension of the science.
I would assert that it would be impossible for someone to create 1000+ string instruments of notable quality without a solid understanding of how these instruments create sound.
If it was just a matter of "copying" one version that by pure luck and circumstance sounded great, then makers should have been able to produce knockoffs of similar quality for many years.
People now know that those great early makers hand tuned the plates until they got distinct sounds. So they were able to reproduce the tone from violin to violin despite variation in density, strength and grain size in the woods.
So today, master craftsman can create violins of comparable excellence. They also have the benefit of modern technology to help them tune the wood and shape the sound in new ways.
This discussion has been archived and is no longer accepting responses.
Violinist.com is made possible by...
Dimitri Musafia, Master Maker of Violin and Viola Cases
Thomastik-Infeld's Dynamo Strings
Violinist.com Summer Music Programs Directory
ARIA International Summer Academy
Johnson String Instrument/Carriage House Violins
Laurie's Books
Discover the best of Violinist.com in these collections of editor Laurie Niles' exclusive interviews.
Violinist.com Interviews Volume 1, with introduction by Hilary Hahn
Violinist.com Interviews Volume 2, with introduction by Rachel Barton Pine
May 19, 2014 at 10:32 PM · Dear Harvey: Thanks so much for all the work you did. I haven't been able to hear your recordings since when I went to the tl;dr: web sites, as I think you instructed, I could not find anything submitted under your name. --- In reviewing your first chart I noticed the place most involved with pulsing was at 4.3K using your boxwood tail piece. I am not sure pulsing is necessarily a bad thing. It may work in the brain to create added attention being given to one's listening faculties each time a pulse occurs. If so, this might explain why you heard my tail piece produce more apparent sound than your boxwood tail piece. I also have a question or two about your second presentation of charts in which their data, when looking at your set of three charts, is not completely clear to me. I checked out my violin with a decibel meter right up to an inch away from the "F" hole and the sound measured 110 dB. I then moved it away by 2 feet and it measured 94 decibels. I think this might be a three times larger result than that which you reported but I don't know how far away from the violin you had the pickup of your meter? In trying to notice some differences between the three charts I assumed the vertical scale represented dB readings. I must ask why then do the values go down as they ascend the axis? Never the less, I will assume if the peaks go higher it indicates a better condition than if they read lower on the vertical scale. In that case I notice that your boxwood tail piece shows 8 peaks ending at over 81 on the vertical scale while my tail piece without foil shows 10 peaks over 81 on the scale and my tail piece with foil shows 13 peaks over the 81 value on the vertical scale. I suppose that shows my tail piece providing more power to the note? In like fashion at 66 on the scale your boxwood gives 14 peaks over 66 while my tail piece with or without foil gives 17 peaks over 66 on that vertical scale. Other than these questions about your observations I can again thank you for your kindness and help in this experiment. John