Quote:
Originally Posted by sumone
So basically you're saying in order to "hear" all frequencies at the same level, the input signal will NOT have all frequencies at the same level?
|
Let me try to explain this from the beginning. We will be simplifying a few things and misusing a few terms here but the essense is the same.
Imagine a sound processor outputting a pink noise via speakers which means equal energy to each frequency. Then we have a microphone pick up that pink noise. On a frequency-amplitude graph we would now see a flat graph.
Now, let's use the mic to record the pink noise on CD and play the CD back using a cd player and speakers (let's say both are Bose products). If we now use a mic again to pick up the playback from these Bose speakers, it would be a perfectly flat graph again, ideally speaking.
So now let's put the mic inside our ear, at the entrace of the ear canal and play back the recorded pink noise via the bose speakers. What happens is that the graph will not be flat because the sound, before reaching the entrace of the ear canal where the mic is located, will interact with our outer ears. The sound has changed and therefore the graph is not a flat line.
So that means that a flat signal doesn't measure flat at the point where it reaches our ear where it matters.
Let me try explain using math functions again: Let's call the output sound signal (outputted via speakers or headphones, doesn't matter) as "x" and what we measure at our mic as "y".
If we place the mic in a room, there's no function because nothing alters the sound or rather, the function is identity ID. So that means: ID(x) = x by definition.
So if we place our mic into the room it'd be ID(x) = y and because ID (x) = x, y = x. So if we want to receive a flat signal y, we have to output an actual flat signal x from the speakers.
Now we place the mic into our ear. Imagine the change of sound via the outer ears as a function again. It's called the head related transfer function (HRTF).
So it's like HRTF(x) = y, but HRTF(x) =! x ("=!" means "is not") by definition (because the outer ear changes the sound)
Therefore y =! x. Therefore we don't measure a flat signal at the entrace of the ear canal where the mic is placed at.
Now imagine two measurements: First, a speaker outputting a flat signal and a mic placed inside our ear. The measured graph is not flat as we learned from above but we hear it as "flat" (because it's been like that from birth
). The second measurement is done by outputting a signal via headphones (we don't know if it's flat or not) and again placing a mic inside our ear at the same spot.
Now! We want both measurements to be the same because that would mean that the headphone has managed to "convince" our ears that the signal is the same as the one outputted by the speakers, that is "flat" in our perception.
But the thing with headphones is that the transducers are so close to the ear, that the sound coming from headphone doesn't stand on its own but automatically starts interacting with our ears the moment the sound leaves the transducers. Therefore, in order for our brain to perceive a signal as flat,
the headphone must not output the same signal as the speakers did because the working environment is different. With headphones you have resonance effects and other funny things. With speakers you have room effects, so that's not really fun either but less significant.
Summing up, we have two issues here:
First of all, the Headroom measurements are done via a dummy head with mics placed inside the ear so even if a speaker was outputting a flat signal, it wouldn't be flat on the Headroom graph (since that mic is placed inside our ear and will be affected by the HRTF). And ideally, a headphone should measure the same so that we hear the same things via headphones and speakers (things like soundstage aside). Therefore, a graph that was taken using a mic placed inside a ear (dummy or real head) -- regardless whether it's a measurement coming from headphones or speakers -- should
not be flat!
A flat graph here would mean that the original signal was not actually flat meaning the speakers must have outputted something really funny. As explained in above, any sound reaching our inner ear will pass through the HRTF function. HRTF (flat input signal) = 'flat output signal' is not possible because:
HRTF (input signal) =! 'output signal', therefore 'flat input signal' =! 'flat output signal', and flat =! flat is a contradiction.
The second issue is that if you were to measure a headphone signal using a mic that is
not placed inside the ear but simply in front of the headphone, it still wouldn't be flat because
the headphone is designed so that it measures the same as a pair of speakers would measure at the inner ear (because we already know that the headphone must not output the same signal as the speakers did because the working environment is different). That's basically the essence of it. It is actually completely irrelevant how the headphone measures when not placing the mic inside the ear.
And besides: Because you cannot isolate and decouple the headphone from the ear/head system, it is impossible to talk about absolute neutrality from a theoretical point of view when talking about headphones. However, it's not like everybody is hearing completely differently -- in fact we pretty much hear the same in the bass and lower/mid midrange regions -- so there are common denominators. And I think that people are blowing the "everybody has different ears, so we hear differently" out of proportions sometimes although at other times some people are omitting this scientifical fact when they shouldn't, especially people claiming that something is absolutely neutral beyond any possible doubt.
I hope that made sense now.
Quote:
| How can any equipment (not excluding headphones) be true audiophile stuff if they don't produce a flat signal (according to our ears of course)? That's what I thought audiophile meant; everything (patch signal, amp'd signal, and speaker/headphone signal) is flat & accurate. |
Good question. The unfortunate fact is that every manufacturer of loudspeakers and cd players and amps, etc. will tell you that only their gear is true to the recording. Certainly, they're all very close but the final permille can only be determined by whatever audio philosophy and approach you believe in, and of course whatever sounds the most natural or most accurate or whatever or simply the best to you. The fact is that the audiophile stuff all start to sound very similar from a certain level on for good reasons. Further, frequency response is only one small part of the story of sound reproduction. You have other issues like soundstage, timing, transient response, imaging, etc. And these are also part of the discussion of what is really "true to the recording". So "flat" is only a part of "accurate". In any case, a violin will still sound like a violin on most speakers since they're all pretty accurate and neutral, but it will not sound the same and that makes the small but nonetheless discernable difference in sound reproduction which gives so many audiophiles something to argue about, and quite intensely on top of that. The differences are sometimes so significant, that some people don't even hear it... Well, sometimes there really is nothing to hear. Such idiotic tweaks like "quantum purifiers" for cables wouldn't exist if audiophile didn't already live on the borderline between sane and insane.
On the issue of neutrality, you can read the article "Road to Audio Hell" by Audio Note. It is a very interesting approach and one that makes a lot of sense in my opinion.
