[John Casler]

I like to think of myself as a skeptical objective, subjectivist.

That is, I most always look for the scientific explanation for something, but generally without such, accept sensory input at face value.

So it is interesting that objectivists and subjectivists are always "pushing" on each other.

To that end, I recently read that an Acoustician and very accepted scientist, stated that if it can be heard, it can be measured, and not only that, it could be measured more accurately than it could be heard by the ear/brain.

This made me wonder.   I could really care less about the "argument", but more about what "can and cannot" be, not only measured, but more importantly, "identified".

It would seem that this is where the arguments "collide".

More directly, I have no doubt, that I could give Bell Labs a CD of Little Fugue in G Minor, and they could measure amplitude, frequency, dynamic range, phase, as well as any number of measurable elements to the piece.

BUT....

Could anyone point to a place on the graph and tell me what different instruments were playing, which notes and sounds?  For that matter, if they didn't "listen" to it, could they "identify" it as "Little Fugue", simply by looking at the measurments?

I doubt it.

It must be like measuring a brain wave and trying to tell what someone is thinking.  Can't be done.   Lying? yes, getting turned on?, yes, getting agry or agitated?, yes, but not what they are thinking.

Could they tell me where thevarious instruments were placed in the soundfield?

Could they tell me the melody, or hum it to me from looking at the graphs or a Mellissa, or Waterfall?

Of course I'm aware of the software that can take a single instrument (keyboard) and write notes as you play them, but our ears can hear a whole symphony, and know when it is a violin, a cello, a trumpet, a trombone, a flute, or a piccolo, a bassoon, or an oboe.

Can a measurment device do this, when all the instruments are playing together in a symphonic movement?

It would seem that while we might all agree that there are devices to measure well beyond the perception of human hearing, that no machine can still "hear" what is being played.

Am I incorrect in this belief?

[JoshK]

If you have the spectrum you could conceiveably tell me what instrument was playing.  When you start overlaying a lot of different instruments I think it would become quite complicated, but a good neural net or other AI technique might make for a decent decipherer.  Its somewhere along the lines of voice recognition software.  

However, all this is beside the point.  Measurements are only as good as how we use them.  Standard speaker (or room) measurements won't tell you if it sounds good but it will tell you where you might have problems.  Its a lot like what I do for a living, its part science and a lot of art.

[Ethan Winer]

John,

> I recently read that an Acoustician and very accepted scientist, stated that if it can be heard, it can be measured, and not only that, it could be measured more accurately than it could be heard. <

Indeed. We can easily measure distortion and other artifacts to more than 120 dB below peak level. But the ear can't get much past 60 dB, and usually a lot less due to masking.

> could they "identify" it as "Little Fugue", simply by looking at the measurments? <

I agree with Josh - this is irrelevant and has nothing to do with the ability of "science" to assess the accuracy of audio equipment. And that is the real issue, where golden ears proclaim they can hear subtle nuances that science doesn't yet know how to measure.

What you are describing is human perception of melody and instrument timbre and stereo placement etc. Localization cues are highly dependent on reflections captured by the microphones, but that has nothing to do with whether ears are better than test gear. Ears vary all over the place, even from one instant to the next, making them a very poor judge of audio quality. Or put another way, when a recording sounds different today than it did yesterday it's your ears/brain that changed, not "cable break-in" or a solid state amp "warming up" or the application of a green felt marker, or any of the dozens of other audiophile theories that defy all that is known by real scientists.

--Ethan

[John Casler]

 John,

I agree with Josh - this is irrelevant and has nothing to do with the ability of "science" to assess the accuracy of audio equipment. And that is the real issue, where golden ears proclaim they can hear subtle nuances that science doesn't yet know how to measure. ...

Hi Ethan,

If you look at the title of this thread, I didn't state anything about the accuracy of measuring audio equipment, so that is not the issue I raised.

I simply find it interesting that some have stated if it can be "heard", it can be measured, as if a measurement can tell us what we are hearing.

We can "hear" the differences between instruments, yet I am not sure that can be measured and displayed as what we hear.

I find that "relevant" to the hobby and what measurments might be meaningful.

And I certainly would not dispute that someday all could be measured and displayed as a graph or chart.

What you are describing is human perception of melody and instrument timbre and stereo placement etc. Localization cues are highly dependent on reflections captured by the microphones, but that has nothing to do with whether ears are better than test gear. Ears vary all over the place, even from one instant to the next, making them a very poor judge of audio quality. Or put another way, when a recording sounds different today than it did yesterday it's your ears/brain that changed, not "cable break-in" or a solid state amp "warming up" or the application of a green felt marker, or any of the dozens of other audiophile theories that defy all that is known by real scientists.
...

I am not stating that ears "are better" than test gear.

However since this whole "hobby/science" is based on the music we hear, then it would stand to reason that, "that perception" would be ultimatly what we would want to measure accurately.

I am not interested in "green markers", cable break in and such.  I was simply wondering if anyone knew if today's test equipment was currently of a sophistication that had moved toward really "hearing" rather than just measuring sound.

Hearing, is the recognition of all the sonic elements and being able to perceive what they combine to create (or more accuratley re-create).

The obs vs subs, seems rather pointless to me (since I consider myself a equal mixture of each) since they are on two different levels of sensation to perception.

That is more what I was getting at.

[SWG255]

It can be reproduced, manufactured and improved upon. Many audiophiles will take the folowing as heresy, but I think we're on the brink of a new revolution in high-end audio. I know that professional recording engineers and musicians have been modeling different instrument and instrument amplifier sounds for awhile now, including models to make the sound recorded by one make of microphone "sound" like another. The obvious benefit of this for the recording engineer or home enthusiast is to make a vocal track recorded with a Shure SM-58 sound like it was recorded with a Neumann U-47. Let's suspend discussion of whether this works well or not and consider the computer modeling and software that makes this possible. It won't work at all unless someone has made a reasonable effort to measure the characteristics of both microphones, so the recorded waveforms from the Shure can be transformed into something that sounds like the Neumann. I believe this is the forefront of something which will begin to appear in audio for general music reproduction. Bob Carver was a pioneer in this effort when he designed amplifier circuits that could mimmick any number of "classic" amplifier designs. Some laughed at his efforts, but his attempts will soon be seen as primitive compared to the digital modeling and transformations that may soon be available.

Will this technology put all the big dollar high-end manufacturers out of business? I don't think so, but i do think that such modeling coupled with loudspeaker/room interaction compensation will vastly improve the sound of reproduced recordings in many audiophile's listening rooms for less than top-dollar prices.

One aspect of this which I wish the high-end industry would research is why many audiophiles prefer the sound of vinyl playback to CD or hi-res digital playback, and if using modeling techniques similar to those for microphone modeling, we could get digital playback that provides consistent emotional satisfaction to even the most die-hard vinyl lovers, with of course, no pops, cliks or groove noise.

I guess this is a long-winded way of suggesting that the question you raise is important, and given where audio playback may be going in the coming years, the understanding of human perception of music playback and the ability to measure key aspects of what makes musical playback more or less enjoyable, should lead to very interesting listening.

[jules]

There is a difference between being able to detect and measure all the component elements of a sound as against being able to record, process and play back the same thing.

I'd like to see more of the original theory but the conditions allowing for These measurement are probably going to require a laboratory situation with a limited amount of sound to be processed, perhaps just a single instrument surrounded by pick-ups.

I think it's laughable to suggest that it might be possible to meaningfully measure all the sound information generated by say an orchestra. [Was this the suggestion John?] The problem is sheer volume and an appropriate analogy might be to say that it's like the challenge of storing all the information in a human brain, in a computer. Technically it's still way beyond us.

It's also laughable to suggest that sound engineers are on the edge of a breakthrough to "improve" on the sounds made by a variety of beautifully made and incredibly complex musical instruments. They are not.

Buy an  instrument, have a live musical evening in your own home, listen to a trumpet in your living room, sit in the middle of an orchestra for a session. Get real!!!!

jules

[Ethan Winer]

John,

> I simply find it interesting that some have stated if it can be "heard", it can be measured, as if a measurement can tell us what we are hearing. <

Those are different things because hearing includes perception. But why couldn't anything you can hear be measured? The challenge is in analyzing. For example:

> We can "hear" the differences between instruments, yet I am not sure that can be measured and displayed as what we hear. <

I'm sure you could use an FFT to measure the spectrum of a Wave file and how it changes over time, and tell if a note is played by a piano or trumpet or clarinet. Even in the context of an entire mix I'm sure an algorithm could separate out the components. Maybe not perfectly, but certainly to a much higher resolution than ears because ears are susceptible to masking and test gear is not.

--Ethan

[Ethan Winer]

SWG,

> models to make the sound recorded by one make of microphone "sound" like another. <

In this case I'd say that "modeling" is simply a buzz word intended to make what is basically the application of a single EQ curve seem more impressive than it really is. You are correct that someone has to measure a bunch of microphones in order to know how to make one have a response similar to another. But there are many reasons these devices are not really "modeling" anything. For example, a U-47 can capture frequencies far above 20 KHz, but if an SM-57 rolls off sharply above 12 KHz (guessing, just for example) then no amount of compensation can recreate the content that was lost when the recording was made. Likewise, microphones vary in other ways that cannot be corrected later. Take proximity effect. Unless the algorithm knows how far a singer was from the microphone, it can't know how much LF boost to apply. And, of course, many singers "work" the microphone and change the distance continually. Then there's off-axis response. If you record five singers in a circle around a microphone, the frequency response for some of them might have less high end or whatever. This too cannot be compensated for afterward since all five singers are already "mixed" together.

> why many audiophiles prefer the sound of vinyl playback to CD <

Oh that one is very easy! Distortion can be a pleasing effect. In fact, I just wrote an article for the current issue of Sound on Sound magazine about this very topic. Read it here:

www.soundonsound.com/sos/feb06/articles/soundingoff.htm

--Ethan

[Ethan Winer]

Jules,

> Buy an instrument, have a live musical evening in your own home, listen to a trumpet in your living room, sit in the middle of an orchestra for a session. Get real!!!! <

Indeed. Many communities have an amateur orchestra for dedicated beginners. I urge anyone who considers music very important in their life to participate. Get off that couch RIGHT NOW and go rent a cello!

--Ethan

[sts9fan]

I simply find it interesting that some have stated if it can be "heard", it can be measured, as if a measurement can tell us what we are hearing.

Any sound wave that our ears can pick up can be picked up my a microphone. The part that cannot be duplicated is the processing. Our brains telling us how things sound as function of neccesity. There is no way of knowing if a french horn really sounds like we think it does because there is no way to take our brains out of the picture. In the same regard you can't prove that the color red is really like you think it is and not just how the human brain processes that wavelength of light. Due to this ultimate computer called the brain you cannot say EXACTLY how something will sound to person A compared to person B assuming they have the same good hearing.
Hypothetically a french horn could sound different to A then to B due to the brains processing. The same waves entering the ear do not have to be percived the same.
we all live in perception tunnels and everything in our lives are as we percive them. This does not have to be how they actually are..

[ctviggen]

And in fact that will happen.  Supposedly, being able to hear notes cannot be taught.  You either get it and can hear them or you don't.  Similarly, there are 'super tasters' who have a bazillion taste buds and can tell you that a cilantro leaf fell by accident into your cauldron of soup.  The human body is amazing and not well understood.

That doesn't mean that a computer couldn't be taught to determine what an oboe sounds like and to pick it out from other instruments.

[sts9fan]

That doesn't mean that a computer couldn't be taught to determine what an oboe sounds like and to pick it out from other instruments.

I would say a computer could find the oboe but it can only be taught how the programmer precives it to sound.