Looking at a sound spectrum and thinking about possible improvements for the sound. (Or better leave it as it is?)

[Andreas Preuss]

So, I got of my violins, a del gesu replica, back in my shop, played it and it didn't sound bad. Actually I thought it sounded better than what i remember when it left my shop.

Anyway, after taking quickly a graph, it looked like this

 

What I would like to see, is actually 'more meat' in the region above 2.2kHz. The rest looks pretty good, even and balanced

Another thing was that it got a kind of difficult response on some notes. Not really worrysome but if anything I want to have an even response. Either an equally bad response on all notes or an equally good response on all notes. I don't like it if some notes play like heaven and others like hell.

For a simple description I'd say, the sound is loud and bright, but lacks depth and some texture. 

 

Technical data is 

Bridge height 34mm

String angle 156

Bridge weight 1.9g

Symmetric bridge model

Strings: Jargar superior set

Sound post in normal position

boxwood fittings

I didn't measure the entire weight,

[Andreas Preuss]

After some time to think it over, I decided not to open the violin, but try to improve things with the setup first. This just follows my general rule that unless any possible measure in the setup fails to fix a tonal problem, the body shouldn’t be opened. 
 
What sticks out most is the high bridge. (Nowadays I try to keep it between 29 and 31 mm) And since the string angle is rather a bit steep (small) this just gets as well adjusted by lowering the bridge.
 

Shouldn’t be too much work to pop off the fingerboard take a few shavings, put it back and lower the bridge to adjust. 

I hope I can manage to keep the soundpost exactly in the same position to get a good graph for comparison (before and after) 

[Don Noon]

With the gigantic caveats:

1) These plots are highly dependent on the operator, equipment, and environment

2) What a violin sounds like and what one thinks it should sound like are highly dependent on the listener

Your plot looks very uneven to me.  The CBR output is unusually strong, and the B modes are relatively weak, which might sound deep but missing some core.  The two high peaks at 700 and 940 Hz (F and Bb on the E string) might make those notes jarringly loud in comparison to nearby notes.  There are two fairly deep/broad dips between 1 and 2 kHz, with the upper dip more concerning as it could detract from the "colors" available.  It would be nice to spread out the concentrated power around 2 kHz.  And more power above 2.2 kHz would be nice too.

The only thing I can think of as an adjustment (in addition to a lower, lighter bridge) would be to move the soundpost slightly away from the bridge.  There is usually a resonance around 840 Hz that is highly sensitive to post position, and there's a dip in the response of your violin that might be filled in with this move.

Again, with the above caveats, this is the bowed response of my "reference violin" that is played by Annelle Gregory professionally.  The resonance at 840 Hz sticks out a bit, but the post has been in the same spot since it was made a few years before, and I didn't want to mess with it.  This is one of the most even responst plots I've seen, and definitely an all-around powerful instrument... but not really an "old Italian" sound.  For that, the midrange would have to be weakened, and the upper frequencies stronger.

[Marty Kasprzyk]

On 2/26/2022 at 11:24 AM, Andreas Preuss said:

>

Another thing was that it got a kind of difficult response on some notes. Not really worrysome but if anything I want to have an even response. Either an equally bad response on all notes or an equally good response on all notes. I don't like it if some notes play like heaven and others like hell.

>

Note evenness is important to me too.  I sometimes play a chromatic scale and measure the dB intensity of each note with Audacity. The results are then plotted which vividly shows all the weak and loud notes (the old Saunders Loudness Test). These are permanent records of what happened when I made changes to the instrument or to compare different instruments.

Although this is very helpful for seeing the effects of adjustments I found it was way too time consuming to do routinely. Now I often just bow glissandos from the nut end to way up on the fingerboard (about 2 seconds long) and look at the Audacity recording of the sound on my computer's screen.  This is very quick to do.
 

A jaggedly shape curve shows an uneven loudness response. If I want a permanent record I take a screen shot of the Audacity recording.  Two of these are attached which show the evenness improvement of my viola's C string after some changes were made.

Glissandos can be played on each string and then the frequency response curve (FRC) can be done on all of them together with Audacity. A glissando bowed FRC is a little different from a bridge tap FRC because it includes the string's character and bowing effects. 

[Andreas Preuss]

22 hours ago, Don Noon said:

Your plot looks very uneven to me.

Yes, that looks somehow like a concern, however when played it doesn’t really jump into my ears. For me the evenness is how the max peaks are lined up and if the air resonance AND the bridge hill max line up. Your graph is indeed super even.

There is absolutely no doubt that each maker tries to follow his/ her personal sound ideals. Mine are somehow different from yours, but this doesn’t make one better than the other. It is just a matter of a different artistic taste. 

Actually I think that makers working with the equipment of physics are on the way to define a new sound, which is entirely different from Strad. It’s a sound of high definition, desirable for many modern performers. What I personally see as a shortcoming in this sound is a certain complexity. (One could almost say ‘artificial dirt’)

————————

That’s why I am a kind of obsessed with texture in the sound. Another thing, which I am interested in is what a very good performer explained to me as resistance. I don’t think it is visible in any sound graph and has mostly to do with the feel of the contact between string and bow hair. (Including the stickiness of the rosin used)

 

I am not so much concerned about absolute evenness in the spectrum.  I actually think some weaker notes (if not too extreme) have some musical function in the interpretation of pieces. It could be that, if they are just placed on the right notes, they make the difference between a very good instrument and an instrument with intriguing sound quality. (My personal speculation.) 

So my interpretation of the ‘defects’ are a bit different. Yes the bridge hill resonances could be broader but also frequencies above 4khz could be stronger. That’s where the texture seems to come from. 

 

My experience with the soundpost is that placing it more inside outside changes the sound more than the distance from the bridge. The problem there is only that it can’t be done with one sound post.

 

I was wondering if my single tailgut tailpiece could bring more depth to the entire sound. It helps the pivoting motion of the bridge. 

 

Let’s hear. I finished the lower-pitch today and tomorrow I can make it playable for the next sound check. Should be interesting. 
 

———-

note: I was wondering how scales would sound on your instrument. I would suspect that there is not much difference between a violin-harmonic  scale ( G D A E etc.) and a violin-inharmonic scale (C# G# D# A# etc.) Did Annelle Gregory mention anything in this direction?

[Don Noon]

2 hours ago, Andreas Preuss said:

Another thing, which I am interested in is what a very good performer explained to me as resistance. I don’t think it is visible in any sound graph and has mostly to do with the feel of the contact between string and bow hair.

I would suspect that there is not much difference between a violin-harmonic  scale ( G D A E etc.) and a violin-inharmonic scale (C# G# D# A# etc.) Did Annelle Gregory mention anything in this direction?

I interpret the player term "resistance" as "impedance"... most often thought of as determined only by mass and frequency... but also depends strongly on the coupling of the bridge to the mode in question.  

When pressed for details about tone and playability, the only thing Annelle ever said was "It does what I want it to do."  I think that sums up a player's perspective nicely, but not much help trying to analyze the details.

[scordatura]

3 hours ago, Andreas Preuss said:

My experience with the soundpost is that placing it more inside outside changes the sound more than the distance from the bridge. The problem there is only that it can’t be done with one sound post.

I agree. The main effect of this is tightness due to the cross arch. Tightness over position.

[scordatura]

3 hours ago, Andreas Preuss said:

That’s why I am a kind of obsessed with texture in the sound. Another thing, which I am interested in is what a very good performer explained to me as resistance. I don’t think it is visible in any sound graph and has mostly to do with the feel of the contact between string and bow hair. (Including the stickiness of the rosin used)

My version from another post:

I think that the way an instrument feels/responds is as important as the sound to a performer. It is tactile. Granted different players have different requirements or skill levels. Instinctively I do think there is a correlation between FFT spectra and playability. But can FFT determine how bowing variables respond when a good player is trying to change the sound? Some instruments have a lot of flexibility and others do not. Can a spectra quantify that? I would say it would be very difficult to interpret that from a chart.

[scordatura]

Andreas, I don't mean to be captain obvious here but what about changing to a different set of strings?

[scordatura]

1 hour ago, Don Noon said:

When pressed for details about tone and playability, the only thing Annelle ever said was "It does what I want it to do."  I think that sums up a player's perspective nicely, but not much help trying to analyze the details.

She probably has not played enough instruments or lacks the ability to verbalize the characteristics. That being said she plays at a high level.

[Andreas Preuss]

11 hours ago, scordatura said:

Andreas, I don't mean to be captain obvious here but what about changing to a different set of strings?

Strings are always on my mind. 
 

However, I think one must be very careful to have a systematic approach when it comes to sound evaluations. I have gathered now enough experience to know roughly where to look when I see some shortcomings or defects in the sound. And usually in this game, strings are most of time at the end of the list. 
 

In this special case I am trying to keep the strings at the moment as a constant, and if I get for the panorama view of the sound what I want (or close enough to be happy with it) I will certainly consider strings. Each structural change needs a newly calibrated setup to perform best.

[scordatura]

On 2/28/2022 at 10:53 AM, Marty Kasprzyk said:

Two of these are attached which show the evenness improvement of my viola's C string after some changes were made.

What changes were made?

[Marty Kasprzyk]

12 hours ago, scordatura said:

What changes were made?

I had been experimenting with the "tuned-mass damper" ideas Professor Jim Woodhouse has for controlling wolf notes (see his attached Youtube talk and his website https://euphonics.org  book chapter 9.4.3). 

https://www.youtube.com/watch?v=gTaOtKuWAHA

I made a long thin wood bar with a graphite fiber outer layer to stiffen it.  One end was glued to a small wood block and the other cantilevered end was repeatedly cut shorter to tune the bar's resonance frequency to a wolf note frequency my viola had (440Hz  open A note).  The bar and block was then glued onto the bass side leg of my viola's bridge as seen the below photo.  It worked fine.

The bar split my viola's wolf note causing resonance peak  at 440Hz into two smaller ones which completely eliminated the wolf note effect as was predicted.

This made me realize another tuned-mass damper could be added to split another high resonance peaks the violin had in its frequency response curve.  Other notes(not necessarily wolf notes) that were overly loud could be reduced while other weak ones boosted thus making the notes more even.  It is quite surprising to see how much the lower end of the  frequency response curve can change. 

Although the cantilevered wood bar worked well I found it difficult to finely adjust its resonance frequency and the process is not reversible.  So I've started to use 2mm threaded stainless steel rods instead and by treading on tiny nuts in various positions along the length I could very finely tune its resonance frequency.

 

 

 

[reguz]

I very much like to know how the recordings are made. Which string is played recording these results? Not the G-string I normally do?

REC 22022022.docx

[Don Noon]

11 hours ago, Marty Kasprzyk said:

I had been experimenting with the "tuned-mass damper" ...

I have done that too.  It does work (yay physics). but the analysis is all about steady-state conditions.  With a low-damping oscillator in the system, the dynamics or transients (starting and stopping a note) get a little strange.  Most players don't seem to like that, so I think this "fix" is a last-resort for a wolf that is intolerable.

It's another one of those things that FFT's won't describe well.

[scordatura]

The mass damper concept made me think of how some formula 1 teams were cheating by using a hidden mass damper in their suspension. Until they got caught…

[Marty Kasprzyk]

16 hours ago, Marty Kasprzyk said:

I had been experimenting with the "tuned-mass damper" ideas Professor Jim Woodhouse has for controlling wolf notes (see his attached Youtube talk and his website https://euphonics.org  book chapter 9.4.3). 

https://www.youtube.com/watch?v=gTaOtKuWAHA

>

Although the cantilevered wood bar worked well I found it difficult to finely adjust its resonance frequency and the process is not reversible.  So I've started to use 2mm threaded stainless steel rods instead and by treading on tiny nuts in various positions along the length I could very finely tune its resonance frequency.

 

 

Tuned-mass dampers are commonly used as Jim Woodhouse pointed out for completely or nearly completely eliminating unwanted vibrations.  However in our case we just want to lower a resonance peak a little so it is not objectionable and we still want it to produce some sound.  This means the weight of the damper doesn't have to be very much.  Some amount of viscous or air damping is helpful.  

Attached is a photo of the cantilevered steel rod with nuts used for adjusting the resonance frequency of the damper to match a peak resonance of the instrument.

[David Stiles]

On 2/27/2022 at 3:24 AM, Andreas Preuss said:

Anyway, after taking quickly a graph, it looked like this

 

Hi Andreas, can I ask what you use as a hammer and how hard you tap? 

[Andreas Preuss]

4 hours ago, David Stiles said:

Hi Andreas, can I ask what you use as a hammer and how hard you tap? 

Hi David, I didn’t  use a hammer. 
 

Though the spectra created with a hammer have a much higher precision when repeated, I trust more spectra created by bowing the instrument. To me it is the most realistic situation for evaluating the sound. For better precision I started to play a half note scale 5 times in a row. That’s what you see in the graph. 
 

 

[Michael Darnton]

I haven't followed what others have said, but your graph looks peaky to me. My objective would be to smooth it out.

[Andreas Preuss]

So now I got time to post the latest results. I wasn't really happy after I only lowered the bridge. Playability improved, but the sound was lacking depth and didn't have enough sizzle (overtones) For completeness here is the graph:

 

So i decided to make some more radical changes. I opened the top, reinforced the linings on the upper bout, thinned down the top from 68g to 65g, put a small protection patch for the sound post.

And last not least I tried to increase the long arch of the top by lowering the area around the lower block which was at 33mm pretty high. In order not to diminish the string angle again, I made a higher lower nut.

 

String angle 168

BH 31mm

BW 1.7g (carbon reinforced)

total weight

425g

Just looking on the graph I like those high peaks between 2800 and 3500HZ. @Don NoonTo me this graph looks more uneven than the graph taken before the alterations were done. But actually the violin plays better now. Before the sound was too one dimensional, now it got more taste and richness, maybe not ideal but definitely better than before. 

The last thing I am gonna to try is setting one of my asymmetric bridges on the violin. 

[Marty Kasprzyk]

50 minutes ago, Andreas Preuss said:

So now I got time to post the latest results. I wasn't really happy after I only lowered the bridge. Playability improved, but the sound was lacking depth and didn't have enough sizzle (overtones) For completeness here is the graph:

 

So i decided to make some more radical changes. I opened the top, reinforced the linings on the upper bout, thinned down the top from 68g to 65g, put a small protection patch for the sound post.

And last not least I tried to increase the long arch of the top by lowering the area around the lower block which was at 33mm pretty high. In order not to diminish the string angle again, I made a higher lower nut.

 

String angle 168

BH 31mm

BW 1.7g (carbon reinforced)

total weight

425g

Just looking on the graph I like those high peaks between 2800 and 3500HZ. @Don NoonTo me this graph looks more uneven than the graph taken before the alterations were done. But actually the violin plays better now. Before the sound was too one dimensional, now it got more taste and richness, maybe not ideal but definitely better than before. 

The last thing I am gonna to try is setting one of my asymmetric bridges on the violin. 

If you are trying to get more close together resonance peaks the easiest thing to do is to increase your Audacity recording size from 2048 to something higher: 4096, 8192, or 16384.  It also makes it look like you have less damping.  

[Andreas Preuss]

On 3/5/2022 at 2:51 AM, Marty Kasprzyk said:

If you are trying to get more close together resonance peaks the easiest thing to do is to increase your Audacity recording size from 2048 to something higher: 4096, 8192, or 16384.  It also makes it look like you have less damping.  

Playing around with the recording size makes me always realize that we can manipulate how we see a sound graph. In reality nothing has changed.

now I often go down with the recording size setting to 256. Somehow it looks closer to what we hear. And there I usually look on the region between 2kHz and 8kHz). I think most of the ‘stuff’ players are interested in is in that region and how the contour looks like.

[Don Noon]

19 minutes ago, Andreas Preuss said:

And there I usually look on the region between 2kHz and 8kHz). I think most of the ‘stuff’ players are interested in is in that region and how the contour looks like.

My thoughts are closer to the Dunnwald bands.... where the "good" zone is 1640-4200 Hz (D&E bands), with higer frequencies being harsh and not contributing musically... unless you like crackly, jangly sounds.

[Michael Darnton]

When looking at FFTs of fine old instruments I expect the spectrum to drop like a rock at 6K and stay there.