I recently bought a sound-meter to measure the volume of pianos. Thanks to the long-suffering patience of many dealers, I was able to measure several different grands, uprights, and even my Grandma's Wurlitzer spinet. I measured them with the microphone in front of the keyboard (because it was a consistent location, and it was easy to hold there. Ideally I would have held it by my ear).

Some raw numbers (in dB):
1985 Steinway B: 100
Steinway L Tiffany: 95
Schimmel 7ft: 104
Schimmel upright: 101
Hailun 5'10": 101
Hailun 6'5": 100
Estonia 225: 101
Weber 5'9": 101
Wilhem Steinberg IQ24 upright: 102
Kawai GX5 EP: 98
Kawai GM11 EP: 98
Bosendorfer 185: 98
Bosendorfer Imperial: 102
30 year old Yamaha upright: 98
Wurlitzer spinet: 97


Some things to note:
Before testing, I thought the Bosendorfer imperial would be at least 10 decibels louder; you can feel the sound in your chest when you play it. But it wasn't, it must be more volume at the low frequency?

All these pianos were remarkably close in range. From this I conclude that the primary limitation on piano volume is the something common in all piano actions, maybe how fast the hammer can move?

The Kawais were incredibly consistent, across hugely different models and piano sizes. I double-checked several times because this surprised me so much.

Some of these I probably could have played louder if I really pounded, but I tried to be respectful to the dealers.

Because the decibel scale is logarithmic, a 3dB difference is a doubling of energy.

The sound-meter I used has an accuracy of +-3.


Has anyone else ever tried anything similar? Were you surprised by the results? It would be great to hear about a separate test.

Incidentally, I didn't put the lower range of any piano (because that is harder to measure, and regulation makes a huge difference), but if anyone is interested, they all went down to the 60-70dB range.

How many fffs would you say you were playing at? Was it an average over the time of playing some piece, or just single notes or chords? Were all the rooms about the same size? This is an interesting subject, I'm just not sure how to standardize a test like that.
I wonder if the common factor was how loud you dared to play the pianos, lest the dealers complain.

Good questions.

Loud as I reasonably could.

On all pianos I played the last page of Beethoven-Liszt Symphony 7 movement 1, a lot of chords, fff. The meter recorded the maximum volume that was reached.


I wondered about room size, and so I tested a piano in a small resonant room, and outside of the room. It didn't seem to make a difference. If you think about how sound dissipates through space, it makes sense that the sound bouncing off the wall would be an order of magnitude smaller than the initial impulse, and thus wouldn't have any effect on the readings (although it can sound muddy).


Possible, but I don't think it's the primary difference because there is no correlation between dB and piano dealer. The Wilhelm Steinberg and the Kawai were at the same dealership, for example.

Another interesting thing (it was interesting to me), if someone was talking in the room while I was measuring, it made absolutely no difference. Humans talk around 75dB, so even if you have several of them talking at the same time as the piano it will have no measurable impact on the volume in the room.

Greetings,
For these sound pressure levels to make sense, we have to know what frequencies you were measuring. If, for instance, you were only measuring frequencies between 27 and 200, I think the bigger pianos would easily out-distance the smaller ones. But if you are measuring between, say, 4000 and 4100 Hz, the smallest piano would be almost the same as the biggest.

"Loudness" is usually referring to our perception of sound, not just the pressure levels of the sound. This skews things, since we are more sensitive to higher frequencies. "Volume" is more indicative of how much actual energy there is in the signal.
Regards,

Something I'm confused about: people will often say that a large piano is "too loud" for a certain space. And the general view is that pianos should be suited to the room. But what I'm missing is why, given the fact that we can put more or less the same energy into the action mechanism, a large piano actually generates a "larger" sound? Where is the energy going in a softer piano? Conservation of energy tells me that energy can be lost through friction and certain other inefficiencies, but why is that different on a large piano compared to a smaller one? If an action is reasonably efficient in piano, regardless of the size of that piano, why isn't a similar energy transferred to the strings, and thus a similar volume given back? Is it that larger pianos have heavier hammers? (but if that's the case, don't heavier hammers compromise the inertia of the action?) Are the longer strings and larger soundboard experiencing less energy loss? Or is it a psychoacoustic phenomenon which perceives a piano with better bass response as louder?

What's going on here? Any thoughts, people?

Were you using A-scale or C-scale? Time weighted average?

I have a problem with your method of playing to produce the sound. Every keyboard has a different touch, and on top of that, how do you know you were not affected by a subconscious feedback loop from ear/brain to fingers? The inaccuracies can affect the results considerably. Maybe you were really testing your ear sensitivity or the loudest note on each piano, which may not have been the same note on each piano. Kind of makes the results hard to interpret.

Interesting!

Was this measured with the lid fully open?

I somewhere read that there is 110dB close to the sound board but this is much more relevant. 100dB is very dangerous for the ears, same as a jack hammer at 1 meter distance!! Did you wear ear plugs?.

http://www.noisehelp.com/noise-dose.html

We are in the minutes-range for safe exposure times

My first guess it's mainly the energy at lower frequencies that matters for the bigger pianos. Say below 200Hz is my quick estimation.

Some of the energy in smaller pianos goes into noise. That will still register as volume, but not as music.

Yes, the lid was fully open. And yes, if you hold the sound-meter by the side of the piano, it gets a lot louder, but it's also hard to get a consistent positioning. Differences of a foot or two can make a huge difference, surprisingly.

Fortunately, we don't play the piano at this volume most of the time; usually it's in the ~80dB range, getting to the top in loud sections.

By all means, please disagree! but do so with your own data.

I spent time on two different pianos calibrating my touch with the sound-meter before I took any measurements (in addition to decades of calibration against my ear). I would love to see someone else try a similar experiment as a comparison. I don't consider these numbers definitive, I consider it a first attempt to collect data.

Incidentally, all pianos seem to have a kind of maximum, where hitting harder has diminishing returns. That is the point I aimed for while playing. I could have really pounded with my fists, but it would have only added a few more decibels.

Thanks for the additional info. So 100dB with the lid fully open and playing fff it is.

Any idea what the value is around with lid fully closed? To my ears it makes a lot of difference, maybe 15 dB lower

>Differences of a foot or two can make a huge difference, surprisingly.

I'm not so surprised. If you are at the front of the keyboard, you are ony 1 feet or so from the sound board. So backing off an additional feed would half the sound volume. Maybe the formula is not holding exactly at such close range (as piano is not a point source), but somewhere in that range.

I once did a unscientific test on my Schiedmayer with an iPhone app. As I recall the reading went up to 100 dB and my ears rang for quite a while.

As BDB says noise can account for a lot of acoustic energy. Measurements on the power sprectrum of harder hammers show that. That may be one reason why musician's earplugs roll off at frequencies above 6-8 kHz.

What model of sound meter do you have?

Not at all surprising. For one thing you're in an area where many resonances can be found and moving a microphone just a little one way or another can make a big difference in sound intensity. The instrument you are using does not indicate sound intensity relative to frequency so you have no way of knowing just what it is picking up.

As well, radiated sound obeys the inverse square law, that is, sound intensity decreases rapidly as the distance from the source increases. Ignoring reflected sound, if you measure a sound intensity of I at a distance of 1m you will find that sound intensity has decreased by I/4 at 2m and I/9 at 3m, etc.

ddf

Much has already been written on this topic, but here is the short version. Whether a piano is perceived as “too loud” or “too weak” is not necessarily a function of it physical size. Some short pianos can sound overly loud in a large room while some longer pianos will still sound weak even when placed in relatively small rooms. Many factors such as the stringing scale, soundboard design and hammer characteristics all have a greater impact on the perceived “loudness” of a piano than does its physical size.

Our perception of a piano’s “power” (or “voice”) is formed within the first few milliseconds at and immediately following hammer impact. We make our perceptual judgment based on rather disparate sound components such as hammer knock (the percussive sounds of impact) and the initial harmonic waveform that is heavily loaded with sound energy developed by the shorter string partials and the initial rate of decay. In addition to hammer knock, the initial waveform is made up of sound waves that can be heavily weighted toward the higher harmonics. In general, the more energy found in the higher harmonics of that initial piano tone the “brighter” and “louder” the piano will sound.

Over the past several decades there has been an emphasis on making pianos louder. Whether this has been done in an attempt to better compete against electronic instruments or against each other is the subject for another paper (and writer) but, thankfully, the industry is finally trending away from this stupidity and once again trying to develop more musical pianos.

Attempting to measure pianos in terms of the SPLs they can generate is an interesting pastime but one that is not altogether useful in terms of piano research. At least not unless it is done under highly controlled conditions and in a reasonably non-reflective environment. Even then, unless you are structuring your testing to record and analyze the waveforms so the energy content of the individual harmonics can be seen and measured the results won’t be of much practical use.

ddf

If that. What you are coming up against is called "action saturation." It is that point at which the keys bottom out against the front rail punchings before the hammers start to move. Beyond this point no matter how much harder you hit the keys the hammers will simply not go any faster.

ddf

There are a few of us who have concert grands at home. It works fine. They're only loud if you play them that way.

The muscle in to music out function is highly variable. My BIL's Daewoo console is louder for the same input than my Knabe in the mf - f region. The difference is that little pianos overload and distort as you go louder, while the big ones have the headroom to stay musical.

My overly simplistic "lay" understanding is that the larger pianos generally have higher efficiency in the conversion of the finger's kinetic energy to sound energy - since soundboards are more efficient in the larger sizes.
As to why that is... it probably just doesn't scale well.
Perhaps some relationships between -- area, thickness, stiffness, weight, resonances and a whole lot of other things.
Oh yeah, coupling between the strings and the soundboard, I can easily imagine that to not scale in a linear fashion.
Any energy lost (not transferred from strings to soundboard) there would lead to a quieter piano.

Hi phantomFive,

Interesting post.

First, to get a feel for the certainty of your readings did you take, say, 10 readings on one piano and calculate the average and the standard deviation? If you did then could you show those numbers? The standard deviation is important because the variation associated with a reading on a piano might be greater than the variation among the pianos which would color the interpretation.

Second, the following two graphs show some data I took on a grand piano, not for the reasons you took yours, but they may shed some light on the problem. These graphs show the sound envelopes for each key of the piano. The first graph plots the strength in linear units. Note in the first graph how the sustain time and amplitude decrease with increasing key number. I went to some detail to make sure each key was struck with the same force. The details of how I did it will appear later this year in a publication.

These envelopes might be used to compare the strengths or loudness of various piano and perhaps augment your data.

The second graph is the same as the first except the strength axis is plotted logarithmically so as mimic the behavior of decibels.

Both graphs perhaps present a different aspect on comparing pianos from the loudness point of view.

That is fascinating.

That's some good, similar data! It actually reminds of the action data I've seen collected at Fandrich piano, although theirs is from a weight perspective. I would be interested in seeing similar data from a Steingraeber.

I did take up to 10 readings of some pianos, but I didn't write them down. It varied from piano to piano. The Estonia ended up with all the measurements within 3dB of each other, but the Kawais (on two very different pianos!) were within 1 dB of each other each time I tried. My own estimate is to not consider anything within 3dB to be significant.

Overall the thing that fascinates me most is how close all of them are.

Also, +1 props to you for supporting Open Source Octave on your website.

I used the Digital Instruments SL-814

One thing I would really like to see is how volume increases as force on the key increases. Is it a linear increase, or erratic?

One thing you might try is to see how the read out varies from note to note played with the the same force.

I can relate this a little bit to golf clubs.
There is a point at which making the club head heavier makes it slower - there being some limit to the amount of energy that a golfer can provide.
Speed matters more than mass, its that M V squared thing.

Just thinking that hammer head weight has been empirically derived and apparently so has the maximum head speed (as some function of the action) - so efficiency of the soundboard and string/soundboard coupling may be some of the few variables available.

The original post doesn't list any Yamaha's which is the brand I measured a few years back to have the strongest, loudest db readings. I offer kudos to Yamaha's engineering staff. I also remember Bosendorfers easily reaching well over 100 dbs.

Good technique on a good piano will produce both a full forte ffff and a pppp that reaches the back of the hall. imho.

Enjoy!

Does this mean projection does not depend on volume, as measured in decibels? What techniques project sound from a piano?

The room acoustics...

Sorry phantom, but the sound level meter (SLM) you said you're using is not an appropriate device for measurements. Consider it a toy. It's the equivalent of a gooseneck tuning lever or a Casio keyboard.

1) The error on this instrument of plus or minus 3dB is an error of 50% either way. It's not nearly good enough a tool to make comparisons.
2)This instrument is not able to be calibrated.
3)This instrument has no octave or 1/3 octave capability - something which would be useful to show the differences between timbres of the different instruments at various amplitude levels. (But even if the SLM did have this capability, I'm still not sure it'd be good enough).
4)You provided no control for the spaces that the pianos were in - the room acoustics can make a tremendous difference in the measured sound pressure level. The reverberant sound level buildup could probably contribute as much as 5 decibels to a reading such as this in an uncontrolled environment.
5)Lastly, you didn't control for the downward force on the keys during the playing of all these instruments. I'm sure you're a fine pianist, but to create a control for the loudness comparisons of pianos, all the pianos have to be played in exactly the same way. I'm not sure, but a portable player mechanism might provide enough control.

To make a responsible comparison like the one you're trying to make, with adequate controls, you have to at least test each piano in the same space, in the same performance location, playing the same source signal, measured at the same receiver location.

It is better to curse the darkness than to light a single candle, I guess!

No, projection is not dependant on volume: projection is roughly based on partial distribution (i.e., brighter/more-focused colours tend to project out in the concert hall).

I want to address these two points specifically:

The goal was not to measure volume at the same force, it was to find the maximum (practical) volume. If the maximum takes more force on some pianos than on others, that is fine. As mentioned earlier, there is a point most pianos hit where further increasing force has diminishing returns in volume.

Secondly, believe it or not, the performance location makes little difference. As Del correctly pointed out, sound power diminishes following the inverse square law, so the reverberations are insignificant. It is surprising but I measured the effect on the same piano in a smaller room and a larger room, just to be sure (because sometimes physics can trick you!). And of course there can be specially crafted rooms that can do strange things with sound, but those aren't the norm....

Yes, it is not surprising that you measured a Bosendorfer reaching well over 100 dBs, depending on the measuring location. Up by the side of the piano, the volume is noticeably higher, so if you hold the meter there, it will give you a higher number.

More exactly, diminishing returns in the increase of volume. Hitting a key with more force will not decrease the volume, except if something breaks.

100 dB in a home or apartment is overkill IMO

Rule of thumb is that repeated exposure to sounds at or above 85 decibels can cause hearing loss. So, all those pianos require earplugs for safety purposes, I suppose.

About the inverse square law. It's a rule that applies only when the distance from the source is great enough that there is a spherical wavefront; that is, when the source looks like a point. It doesn't apply close to a source, or in a reverberant environment. Reverberation can add a lot to the build-up of sound.

You won't get that far from a piano soundboard, and it won't be suspended in empty space, either.

Still, regardless how far the methodology here is from the ideal, it's interesting data, especially the discovery that there's a saturation level for each piano.

Thanks. I'm glad you find it interesting.