Spectrograms...

Lalz

Lalz

Well-known member
Joined
Mar 30, 2011
Messages
1,401
Reaction score
3
Location
UK
...Why don't we use them to illustrate how different types of wood sound?

I know that no two ukes are the same but this technique is used to analyse things as complex as the human voice, so it would definitely help a lot in terms of describing spectral qualities of say koa vs. mahogany vs. spruce woods to the untrained ear. Just using words such as warm, mellow, bright or crisp make it difficult for beginners to grasp what that actually corresponds to in terms of sound, unless one already has experience hearing the differences.

You'd be able to visualise that one is stronger on mid-lows, another one on highs hences brighter, one has more sustain etc. If someone made a reference table of spectrograms of various types of standard woods in identical lab conditions or if makers made "for reference only" ones for each of their models, it would probably help people get an idea of the differences among the various woods. Even better, one could add a vibrometer scan to visualise how the uke resonates to say G, C, E and A.

Do you reckon?
 
Actually, the human ear is capable of hearing more subtle differences in the character of sounds than can reasonably be displayed by a spectrum analyzer (unless there are some incredibly capable analyzers far more capable than the stuff I've worked with).

A trained human ear can hear down to 1-cent or less across approximately 20hz to 18 or 20khz. Getting that degree of precision out of hardware or FFT software would require a huge FFT matrix and very fast sampling - even today's computers wouldn't be able to process it in anywhere near real time - or even reasonable time. :)

In fact, I'm not sure you could reasonably display the information in a visual format that would be meaningful.

John
 
True that the human ear is much more precise when you're listening to the uke in the flesh, there's nothing better than trying a uke directly. But when you can't and you have to listen to comparisons or reviews on youtube for example, you hear the sound through a long chain of audio devices and acoustic factors that will deteriorate the sound and affect how you hear it. You won't be experiencing the intensity nor directional properties of the sound it projects either etc.

With spectrograms and vibrometer scans you get reliable and immediate images of the frequency response through time to a stimuli, which can nicely complement an audio recording, and with very little equipment needed. People use them all the time to analyse other types of music instrument acoustics and other audio signals, it's an established method. Many years ago I worked in an acoustic research lab that used them a lot, although I wasn't working with them myself (I've changed field since then). I remembered about it today when I was trying to hear the difference between ukes made of different woods on youtube. Kept thinking: "that's nice but if only I could just see a spectrogram / vibrometer scan, I'd get a better sense of the actual acoustic properties of that wood / uke and I'd be able to really compare them objectively"
 
I'm also curious about this, and have a suggestion for how to measure the response.

In the speaker building community, there is a lot of attention paid to response measurement and some useful software tools that could be applied to measuring the response of ukuleles with different build characteristics.

For example, this software provides a variety of spectrum visualization tools and is free: http://www.hometheatershack.com/roomeq/

It's important to isolate room resonances from any measurement, and to use a calibrated mic.

I've also been curious about visualizing the physical vibration patterns on the top of a uke using a strobe light. Like this sort of experiment, but with a uke: http://youtu.be/q4setd7BZWM?t=37m53s

My guess is that there is a lot to be learned about bridge and brace design as well as the general resonance patterns of different tone woods.
 
Hi

I think the most serious problem is that our language (and aural psychology) does not correlate to the frequency spectrum. We say something like "this ukulele sounds warm" but I have not seen a consensus as to what frequency does that to our ears.

In addition, the time domain information (eg, attack, rise and decay etc) will be lost in Fourier transform. The phase information is usually lost (unless you do sine and cosine spectrum analysis).

And to get a really good recording is remarkably hard.

Cheers
Chief
 
Indeed, it is really difficult to describe tonal quality and also very hard to make a faithful recording, especially with non professional equipment. So I salute the effort put into this by all the people who are doing online videos and thank them for sharing their super valuable knowledge with the rest of us. :cheers:

Maybe the manufacturers - who can afford renting anechoic rooms and using pro-level microphones - could use these techniques to add the imagery to their spec list? (I'm just brainstorming out loud hehe)
 
Maybe there's a master's thesis in acoustics waiting to be written about this... "Correlating perceived tonal response of different ukulele woods to their spectral analysis: a mapping of the psychoacoustic perception of warmth and brightness in small wooden string music instruments" lol
 
Lalou said:
Maybe there's a master's thesis in acoustics waiting to be written about this...
Click to expand...

Well, I bet there have been many many attempts. This sure sounds like a graveyard for a graduate student, and that for a PhD thesis. And I am sure everyone in the audio/sound/recording industry is working on this.

Cheers
Chief
 
I meant specifically for the ukulele... ;) Haven't seen much work on this area, people in acoustic research tend to work on guitars and violins and such
 
But wouldn't the tonal response of the wood be totally subjective, based on thickness, grain, and bracing used? Not to mention finish, strings, bridge, saddle, nut, style of string attachment, and method of attack (plucking vs. strumming, nails vs. flesh)?

Maybe that's why no one has quantified anything yet - there are too many variables, even within a single type of instrument wood. Speakers would be easier, especially if a standard existed for thickness and finish.


-Kurt​
 
ksiegel said:
But wouldn't the tonal response of the wood be totally subjective, based on thickness, grain, and bracing used? Not to mention finish, strings, bridge, saddle, nut, style of string attachment, and method of attack (plucking vs. strumming, nails vs. flesh)?

Maybe that's why no one has quantified anything yet - there are too many variables, even within a single type of instrument wood. Speakers would be easier, especially if a standard existed for thickness and finish.


-Kurt​
Click to expand...

It's done in music acoustics research for other instruments, but not specifically for the ukulele (as far as I know) and it doesn't seem to be done in the uke industry, or at least resulting in data that customers have access to.

Using visualisation such as holographic or vibrometer scans (not sure about the exact terms in English but I think they are the right ones) actually helps determine the influence of construction details such as bracing etc on the tonal quality. And spectrograms help have a better grasp of whether the sound has a lot of overtones or not, how much sustain there is etc. But this has to be done under control conditions.

I just thought it'd be useful if manufacturers started using these as references to give an idea of the tone to customers who purchase ukuleles online. But maybe I'm just nerding out :eek:
 
I thought that Kawika Hurd did some of this work when he was building `ukuleles back in the '90's.
 
Hm, I feel you're sort of missing the point and making it look more complicated than it really is. It's not about becoming a better uke player, the idea is to try just to get a better sense of the differences in tonal quality of a wood in a uke (koa vs spruce vs mahogany etc) for the many people who don't live next door to a big ukulele store and don't have the opportunity to test for themselves before ordering one, or have a hard time hearing the difference on compressed youtube videos. These visualisations are not that hard to read, they're really quite straightforward.
And to do a spectrogram f.ex. you don't ask a person to sit and strum to make the analysis, you stimulate the surface mechanically to get a frequency response and you're done. Not much manpower needed at all. I don't mean that manufacturers would do it for every single uke produced, just that they could maybe do one per model "for reference only" and publish it together with other specs on their website or something, like "here's the tonal response of one of our koa soprano compared to the same model with cedar top". Or that someone could do a wood chart of the different tonal qualities of different woods or something like that. Just brainstorming. Personally, I think it would be useful to me, but it's just my personal opinion. Also it's sometimes done for guitars so why not the uke?
 
And a decent inexpensive professional level violin costs more than a high end uke. There's no money for this kind of research for ukes.

Google: "Kawika" David Hurd (has been mentioned), Roger Siminoff, Carleen Hutchins, Dr. David Cohen, T. D. Rossing, and Al Carruth for starters. They've all done very in-depth work in the field of instrument acoustics.
 
And a decent inexpensive professional level violin costs more than a high end uke. There's no money for this kind of research for ukes.

Google: "Kawika" David Hurd (has been mentioned), Roger Siminoff, Carleen Hutchins, Dr. David Cohen, T. D. Rossing, and Al Carruth for starters. They've all done very in-depth work in the field of instrument acoustics.

This stuff is not as easy to quantify as you'd like for it to be.
 
Thanks for the references. I'm mainly curious about what's out there already. Agreed that guitars and violins are getting much of the attention in this area of research.

As far as quantifying my love for a particular uke sound, I agree that's extremely hard or impossible to do. However, I have noticed some difference between instruments that seem both measurable and useful at least to me. Some instruments have much longer sustain than others, others have dead spots on certain notes, some have weird intonation issues (notes that start sharp and end flat) and others bias brighter or darker in tone. Certainly, there are many variables that impact all of those things such as strings or even humidity. Let alone players. But wouldn't it be interesting to see if a certain bracing is just much more efficient than another across 8 of 10 instruments tested?

In other words... it's hard to know that little can be learned from quantified testing without data.
 
The active builders I referenced, Al Carruth, Dave Cohen, and Roger Siminoff, have used their research to guide them in making instruments with very consistent tonal and amplitude results. They know what they are going for, and they can achieve it.

Not that intuitive builders can't do that, too, but there are real advances in the science taking place. All three of the builders cited have a lot of wood chips under their fingertips, too.

I'm most interested in deflection testing which Roger has found to have a strong correlation to his tap tuning methods using a special hammer and a strobe tuner to literally tune tops, braces and sections of tops. I took his workshop about a year and a half ago. Very interesting, to say the least.
 
OldePhart said:
Actually, the human ear is capable of hearing more subtle differences in the character of sounds than can reasonably be displayed by a spectrum analyzer (unless there are some incredibly capable analyzers far more capable than the stuff I've worked with).

John
Click to expand...
John the stuff is pretty much out there and the military's passive sonars have been using it for some time. They use a modified FFT and a fairly rapid sample rate. Its been to long for me to remember all the details ,but we used a very narow band and searched for specific signatures. But like you implied the broad spectrum used to evaluate an instrument from sub sonic to ultra sonic and apply some evaluaton of resultant sound , damn near impossible.
Oh yea the other problem, it would give existance of the frequency but since it processes into white noise the amplitude and distingtion would really take many different prolonged samples.
 
Last edited:
The original Highlander coaxial pickup that I helped to develop started off as a passive sonar sensor. Hunt for Red October, and all that...
 
You must log in or register to reply here.
Top Bottom