Thought Experiment

[boom3]

Some decades ago, Fletcher-Munson derived a curve that represents the sensitivity of the human ear. That curve has been refined over the years, but still has the same general shape.

http://en.wikipedia.org/wiki/Equal-loudness_contour

is a good general explantion.

So, if these curves are valid (and reading this porttion of the article)

"Equal loudness curves derived using headphones are valid only for side-presentation. Frontal presentation, from a single central loudspeaker, can be expected to show reduced sensitivity to high frequencies, which are partially masked by the head, and presentation using two loudspeakers, as for stereo will reveal more complicated differences related the HRTF (head related transfer function) which is also dependent on elevation of the sources and plays a major role in our ability to locate sounds."

It has always seemed to me that a loudspeaker attempting to have a flat perceived response should have a response curve like the F-M plot, not a flat "frequency response" plot as measured by instruments.

Even folks who've never heard of Flecther-Munson, in the days when every out-a-site stereo rig had a graphic equlizer, set their sliders to the "smile' pattern-boosted on each end and flat, or slightly depressed in the middle. A crude approximation, but in the same line of reasoning. I suspect of a lot of car audio is set up this way, again, without any suspicion about the research behind it. Boom! Boom! Rattle that sheet metal!

I've asked this of several industry veterens. They laugh and say, "Well, you're probably right, but...y'know, the bass power requirements would be enormous at the low end, and, it's too late to reducate the public to this way of thinking". They also point out that "pro audio" -what we used to call PA-doesn't pursue flat response per se, but rather results that sound good for each particular venue.

Thoughts...?

[Mr. RF62]

To me it would depend on what material is playing, where (place) it is being played/listened to(room reverberation), temperature. There is alot of issues that come into play.

[Islander]

The Fletcher-Munson curve is actually a series of curves, varying with different sound levels, to show that human ears' sensitivity to different frequencies varies with the volume level. At higher levels, it approaches a flat response, so that the correction curve is only needed at low volume levels.

Yamaha stereo amps and receivers (and possibly some other brands) have had a variable loudness knob since the '70s. It's set to flat at loud listening levels, then to reduce the volume, the loudness knob is used, while leaving the volume knob at its loud setting. As the loudness knob is turned, the volume is lowered as per the F-M curves, reducing the midrange the most, so as the volume goes down, the frequency response will continue to be perceived as flat.

It's a logical concept and works fairly well, but doesn't seem to have been picked up by many other manufacturers, who usually go with a single loudness setting (with a button) for low-level listening.

[BLSamuel]

I would think that a speaker should have as flat as frequency response as possible.

Using live music as an example, if I hear live music, then what I've heard has been affected by the Head Related Transfer Function (HRTF).

We'll assume the same live music event is recorded. Let's also assume that the sound in my listening room is perfectly flat at the listening position. If I apply the HRTF to the sound produced in my room, then what I'll get is not an accurate reproduction of what I heard at the event.

This scenario would be assuming live listening levels and not lower levels where the ear if my rusty memory is correct is less sensitive to some frequencies in which case the F-M correction would be desired for perceived flat response.

In other words, my opinion would be that F-M equalization should only be applied to retain perceived flat response - doesn't it flatten out at louder listening levels (yeah, I know I should go read up on it ...)?

Just my .002 cents and over simplified but I've often been called simple ...

[PrestonTom]

There are several points of confusion here.

Only the lowest curve is a measure of sensitivity (i.e., is the sound detected or not when there is some minimal amount of energy). This lowest curve will frequently be referenced as a MAP or MAF curve (minimum audible pressure or field).

The curves above the first are are a collection of points where the sound will be detected and it will also have a perceived similarity in its loudness for various frequencies. The different curves are simply referenced to different various reference loudness values. This creates a psychophysical scale of loudness in units of "phons".

For very high frequencies (very small wavelengths) the sound will not easily diffract to the far ear or around the head. Hence, less energy at the ear when its direction is off axis (away from 90 deg which is pointing at the ear). So with speakers in front, as well as a flute player etc in front, the sound (regardless of the source) should show this same degree of diffraction. I won't touch the issue of HRTFs. You mentioned the issue of headphones, what was not mentioned however, was that headphones (outside of audiometry) are not designed to produce a flat response at the ear drum. Rather they approximate the sound response at the entrance to the ear canal and include the possible effects of diffraction by employing a correction function. This is referred to as free-field correction or diffuse field correction depending on what original sound field is being approximated. There are actually a number of difficulties measuring this on actual earphones at very high frequencies.

Perhaps this will make it clearer. If you were to boost the frequency response of the highs for a speaker in front of you, then if you turned your head (ear now nearer the speaker) nothing would be appropriately approximated in either case.

Part of the problem is that the WIKI article attempts to put several chapters of material into a 100 word essay and makes a poor attempt in assuming how much background a reader has.

Problem solved,

-Tom

[djk]

(and possibly some other brands)

Before Yamaha, there was McIntosh.

I used to sell both, variable loudness is useful.

The bass control on the APT preamp had the same inflection points as the F/M curve, no treble boost.

The loudness button would be OK if you could adjust the level of the source to match (like my SWTPC preamp did).

[Islander]

This is why most "loudness compensation" controls are linked to the volume control pot.

Even the single-button loudness compensation?

[BLSamuel]

SWTPC - hadn't saw reference to that in a long time....

[pauln]

The F-M curves are what they are, a reflection of how we hear relatively less of the extended frequencies at lower volume. This is the natural way we hear sounds and music.

Unfortunately, the common interpretation of noticing that the curves flatten out at high volume is to infer that the perception of the listener should be "corrected" at all lower levels by boosting the bass and the treble.

Since the F-M curve clearly demonstrates that the ears are not designed to hear it flat at lower volumes, the attempts to do so are misguided and result in un-natural sounding music.

The use of loudness controls, turning up of tone controls, and the "smiley" pattern on graphic EQs is a common way of getting that "HiFi" sound - boom and twinkle. It's hard on the equipment, and get's to be hard on the ears.

[djk]

If you don't listen to music at exactly the same level it was produced at, the tonal balance will be off.

Correct loudness compensation is much better than no compensation.

A shot of the SWTPC preamp with individual input level controls, and loudness. The loudness varies with the main volume control setting, the individual level controls allow for all sources to be the same level and for the loudness to track the F/M cure in a closer fashion.

[boom3]

Everyone, thanks for the comments. Pauln, love that avatar! PrestonTom, I'm not sure I understand your remarks, except to agree that any "correction" for 90 degree listening is not right for 0 degree listening. Agree that the WIKI article is simplistic, but I wanted the discussion to proceed from a common point of departure.

Here's what I think...on the low end, room gain, to a degree, compensates for the roll-off. On the high end, I think that the EQ applied by the recording process compensates to an extent. I also agree that the smiley face of a graphic EQ is not, and never was, a panacea. It was a bit of audio folklore.

Some years ago, I read a discussion of the concept of flat frequency response, as in, trying to maintain that throughout the recording and reproduction chain, and one of the conclusions was that really flat FR in the home was a chimera. Not the least of the reasons was that LPs (the medium then in use, along with tape) just had too much noise in the last octave and trying to have a flat response (again, across the entire recording & reproduction chain) was a bad idea. If Mike Bentz is following this, maybe he can discuss the "house curve" that some studios have. My impression is that the typical house begins to roll-off above 8 KHz or so in a gentle slope and is surprisingly low at 20 KHz. How this gets overridden in the final mix, I don't know. This may be old data from an old audiophile []

I do recall that when LPs and RTR tape were my mediums, I frequently raised the treble, but rarely the bass. These days I either leave the tone controls at flat, or for HT, use the Yamaha's YPAO "measured compensation" feature, which still needed some manual intervention to get channel levels right. I sometimes defeat it, particularly with remastered works to see what the remaster sounds like by itself. I'm trying to say that most CDs are hotter in the treble than the LPs I used to listen to. One would think with aging ears, I'd be raising the treble as time went on, but with CDs that is just not the case.

[Daddy Dee]

djk,

The SWTPC preamp is a cute little rascal. Did you build it? About what year?

[djk]

I built my first piece of SWTPC in 1972 for cash (I built Heath, Dyna, and SWTCP to spec).

That preamp sounded fantastic (with an external supply), but was time consuming to assemble.

[pauln]

If you don't listen to music at exactly the same level it was produced at, the tonal balance will be off.

Off compared to what? Are you assuming the flat response at loud volume to be the norm?

At lower levels you don't normally hear as much of the bass and treble, naturally.

If you were approaching an outdoor concert from a distance, would you think the tonal balance was "off" until you got right up close?

Is there really a "correct" level for listening to a jet engine? A noisy crowd? A ticking clock? Music?

[djk]

I'm sorry you don't get it.

I expect a piano to sound like a piano in a real sized room (like it was recorded in).

If you're happy with the sound of a piano from very far away, fine.