or Connect
Head-Fi.org › Forums › Equipment Forums › Sound Science › Audiophiles - Listening to music or to gear?
New Posts  All Forums:Forum Nav:

Audiophiles - Listening to music or to gear? - Page 3

post #31 of 46
Quote:
Originally Posted by ubs28 View Post
 

 

So people adding tubes in their chain makes their signal sound more low-fi? Also a lot of professional producers add analog gear to their chain because they don't like the sterile clean sound which doesn't have the harmonic distortion. 


I think most likely those tubes are for equipments used in music production, not for mastering or playback..

post #32 of 46
Quote:
Originally Posted by BaTou069 View Post


So what do you think? Are audiophiles listening to music, or to gear?

What happens when the gear-horny audiophile found his perfect gear? Will he stop critical listening and will be able just to enjoy the music?

What are your opinions?





On my home forum, oriented toward extreme performance, I have one of the more modest record collections (there are also a couple of thousand CDs on the other side of the room). Does that answer the question?

Edit: A link to an essay discussing the objective, and the appropriateness (if you will), of critical listening.

http://www.stereophile.com/asweseeit/are_you_a_sharpener_or_a_leveler/index.html
Edited by Shaffer - 12/21/13 at 3:23pm
post #33 of 46
To the OP:

When you listen to music, do you listen to the musician or the sound he makes?
post #34 of 46

@ubs28: There's a difference between music creation and audio reproduction. The very definition of hi-fi means flat frequency response, low distortion etc. at reproduction.

 

 

Quote:

Originally Posted by kn19h7 View Post

 

For most I can only comment based on information I managed to find on their websites and online, at least those brands I listed seem like serious business to me. Nitpicking on already low enough output impedance values doesn't seem very meaningful to me... There are other important metrics like FR, channel balance etc. And at least for BHA-1 your data is very different from mine. I have one and it came with an individual measurement cert, unweighted SNR is about -107~109db for both channels, and THD+N is about 0.001~0.003%. Output impedance is listed as 2ohm per opamp, which means 2ohm for single-ended output and 4ohm for balance output.

Can't we expect <1 ohm output impedance with amplifiers costing way north of € 1,000? Sure if the amp is made to drive 300+ ohm headphones or only costs a fraction then it is nitpicking, otherwise I'm not so sure. Because otherwise I also want to be able to drive 16 ohm multi-driver in-ears that have very funky impedance curves.

 

I got my data from audio/stereoplay.de. Even bryston specifies noise as -103 dBV with inputs shorted. dBV means over 6 dB better numbers compared to the "standard" -4 dBu. Inputs shorted means again a couple of dB better numbers than in reality where you actually have a source connected to the input. So -87 dB doesn't seem unrealistic at all.

As for output impedance, I don't trust manufacturer specs. I trust my source since they've also measured other amps with output impedances as low as 0.4 ohm. I also cross checked some measurements. They measured 23 ohm for the AK100 for example, where other sources report slightly higher numbers. Btw, Astell Kern, that's another big fail. Who wants a portable player that is very likely going to be used with in-ears if it has a 23 ohm output impedance?

 

As for flat FR and tight channel tracking.. I presuppose that in any well built amp. Problems here are K.O. criteria for me (and some amps I've listed above do have problem in those areas too...)

 

 

Quote:
At least to me there are other aspects to consider beside measured performance and I am willing to pay for that.

To me there are too. I just said I'm not the kind of guy that opens up devices to drool over the fancy expensive chips used. If anyone does that he/she just biases him/herself.

 

 

Quote:
I said, LISTENING TEST demands full concentration (Did you ever try? I know I did.), not graph starring... Hardware Benchmarking may take long time, but the actual effort required from human operator is very low and simple. And this whole apples with oranges thing is brought up by yourself...

 

Well actually I don't know what blind tests to do with graphic cards...

Audio is somewhat different that its more analog based, each number/graph measurement tells you different aspects of the performance but not everything. Similar case in computer hardware is perhaps display monitor, where one needs to actually take photos to compare certain performance metrics.

No you didn't understand what I was saying.

In audio we (still) need blind tests, with computer hardware we're already one step forward because people accept measurements there.

 

The counterpart to hardware benchmarking is measuring audio hardware. Here I argued that hardware benchmarking can very well take longer.

 

You would do a graphics card blind test as with anything else: set up two identical computers except for the graphics card, run the same application, switch graphics card by connecting the monitor to the other computer, let the participant decide if he perceived a difference.

With amp blind tests it's exactly the same. You switch the amp by connecting the headphone to the other amp (or by using a switch box) and let the participant tell you if he perceived a difference.

Repeat n times, crunch to numbers to see if there's a high change the participant didn't just guess randomly.

 

 

Quote:

Computer hardware market is filled with fancy expensive rubbish too, so as most if not all commercial markets. Actually I don't quite understand why you are so negative on audio equipments... I don't see less problem in computer hardware industry anyway

Like what? Who buys the rubbish and most importantly: what's the justification?

 

I'm not sure you grasp the advancements that were made in computing. Just $25 buys you a computer nowadays that we couldn't even dream of 10 years ago.


Edited by xnor - 12/21/13 at 8:26pm
post #35 of 46

Did you read my comments regarding monitors? For example, you just can't show properties like motion overshoot in numbers or graphs, only way is to actually take photos for comparison. Most computer hardwares work purely in digital domain or just calculation (where the performance concern is only the time it takes), that's why we can just measure performance in numbers and graphs. But for things involved analog domain like audio and visual, numbers and graphs can't tell you everything, so we have things like FR, CSD, IMD, impulse response, etc. Each tells something about the performance, never everything. So we have listening test as a direct way to get an impression of actual performance as a whole.

 

Well, I guess its cool that you are fine with computer hardware, I really don't know how $25 can buy me a practically useful computer though.. But I don't want to go further OT...


Edited by kn19h7 - 12/21/13 at 9:38pm
post #36 of 46
Quote:
Originally Posted by kn19h7 View Post
 

Did you read my comments regarding monitors? For example, you just can't show properties like motion overshoot in numbers or graphs, only way is to actually take photos for comparison.

There are lots of numbers and graphs for monitors. Why can't you quantify the amount of motion blur / overshoot in terms of numbers? Showing different pictures for different settings is nice but all I want to know is the setting where there's the lowest blur/overshoot.

 

Quote:
we have things like FR, CSD, IMD, impulse response, etc. Each tells something about the performance, never everything. So we have listening test as a direct way to get an impression of actual performance as a whole.

The sum of these things (plus some more) tells the whole story, and there's also null-difference testing.

 

For headphones I kinda agree, you should always have had the headphones on your head before buying them. The best measuring and sounding headphone might still cause your ears to fall off after a couple of minutes. :D 

post #37 of 46
Quote:
Originally Posted by xnor View Post
 

There are lots of numbers and graphs for monitors. Why can't you quantify the amount of motion blur / overshoot in terms of numbers? Showing different pictures for different settings is nice but all I want to know is the setting where there's the lowest blur/overshoot.

 

The sum of these things (plus some more) tells the whole story, and there's also null-difference testing.

 

For headphones I kinda agree, you should always have had the headphones on your head before buying them. The best measuring and sounding headphone might still cause your ears to fall off after a couple of minutes. :D 


...so lets show an number or unit used to quantify motion blur? Knowing which settings to use is apparently completely different thing from measuring performance isn't it? Of course I know we have lots of numbers and graphs for monitors, its a similar case as what we can see in audio: we have various kind of measurements, trying to capture as much information as we can.

 

Well, to you sum of several measurements tells everything of your concern, but to me that's not the case. And I don't really think listen to null-difference-ed result is very meaningful, as it would be very different use case from real usage.


Edited by kn19h7 - 12/21/13 at 11:00pm
post #38 of 46
Quote:
Originally Posted by kn19h7 View Post

 

...so lets show an number or unit used to quantify motion blur?

Yes, or a graph of the temporal function. I see no problem there. The only reason I can think of why pictures are used is because they're simpler to "read" by the average joe.

 

 

Quote:
And I don't really think listen to null-difference-ed result is very meaningful, as it would be very different use case from real usage.

Yeah, but two completely different devices with similar FR, noise floor, harmonic distortion etc. that null deeply shows us that there are no hidden (magical) properties.

If they didn't then we could find out what numbers we're missing.


Edited by xnor - 12/22/13 at 12:35am
post #39 of 46
Quote:
Originally Posted by xnor View Post
 

Yes, or a graph of the temporal function. I see no problem there. The only reason I can think of why pictures are used is because they're simpler to "read" by the average joe.

 

 

Yeah, but two completely different devices with similar FR, noise floor, harmonic distortion etc. that null deeply shows us that there are no hidden (magical) properties.

If they didn't then we could find out what numbers we're missing.


The problem is that the graph will look even more complex than the pictures themselves, and I think most likely its only applicable to the specific test pattern or perhaps even pixels involved. I not sure if I'll call this as quantification, more like a conversion instead.

 

But null difference is a case-by-case thing depend on the test samples..


Edited by kn19h7 - 12/22/13 at 2:21am
post #40 of 46
Quote:
Originally Posted by kn19h7 View Post
 

The problem is that the graph will look even more complex than the pictures themselves, and I think most likely its only applicable to the specific test pattern or perhaps even pixels involved.

Yeah but the objection was that numbers/graphs cannot tell you how blurred etc. the image will be.

As for specific test patterns, I'm not sure if you need to specify that but if you do then you should. It's same for audio measurements. 20 Hz - 20 kHz means nothing but if you add +/- 3 dB then it's not useless anymore.

 

This is no different for subjective reviews btw. You should know the test setup, the music used for the test, and even the listener and his/her preferences. Quite problematic if you want any kind of reproducibility. Less so with clearly specified measurements.

 

 

Quote:
 But null difference is a case-by-case thing depend on the test samples..

Kinda, but no reason to worry. The usual test signal is pink noise since it contains all frequencies with roughly the same energy as real music, but you can also use real music of course. From tests I've seen the pink noise was harder to reproduce than music.

 

D. Self on null difference testing:

Quote:
The most positive proof that Subjectivism is fallacious is given by subtraction testing. This is the devastatingly simple technique of subtracting before-and-after amplifier signals and demonstrating that nothing audibly detectable remains. It transpires that these alleged music-only mechanisms are not even revealed by music, or indeed anything else, and it is clear that the subtraction test has finally shown as non-existent these elusive degradation mechanisms.

He's of course talking about differences that are asserted to be audible to golden ears but not detectable with conventional measurements.

post #41 of 46
Quote:
Originally Posted by xnor View Post
 

Yeah but the objection was that numbers/graphs cannot tell you how blurred etc. the image will be.

As for specific test patterns, I'm not sure if you need to specify that but if you do then you should. It's same for audio measurements. 20 Hz - 20 kHz means nothing but if you add +/- 3 dB then it's not useless anymore.

 

This is no different for subjective reviews btw. You should know the test setup, the music used for the test, and even the listener and his/her preferences. Quite problematic if you want any kind of reproducibility. Less so with clearly specified measurements.

 

 

Kinda, but no reason to worry. The usual test signal is pink noise since it contains all frequencies with roughly the same energy as real music, but you can also use real music of course. From tests I've seen the pink noise was harder to reproduce than music.

 

D. Self on null difference testing:

He's of course talking about differences that are asserted to be audible to golden ears but not detectable with conventional measurements.

 

Subjective reviews mean very very little to me, well perhaps those from certain selective individuals mean slightly more. The most useful subjective review to me will be from myself anyway...

 

Regarding the null difference graphs, I think variation across frequencies and time domain is important too. I would suppose it has to be really flat throughout without up-and-downs to prove identicality.

post #42 of 46
Quote:
Originally Posted by SanjiWatsuki View Post
 

There's a difference between adding 0.1% THD+N and introducing 10% THD+N, though.

 

I think this a clear case of someone trying to listening with his eyes. It sounds far from low-fi.

 

The Shure SE535 beats the Sennheiser HD800 in terms of square wave tests and %THD. So based on these graphs you'd think they belong in the summit fi section if one was to listen with his eyes.


Edited by ubs28 - 12/22/13 at 5:19am
post #43 of 46
Quote:
Originally Posted by ubs28 View Post
 

 

The Shure SE535 beats the Sennheiser HD800 in terms of square wave tests and %THD. So based on these graphs you'd think they belong in the summit fi section if one was to listen with his eyes.


Not really, SE535 has serious phase-coherence problem and it can also be seen in 300Hz square wave graph. (based on graphs at innerfidelity)


Edited by kn19h7 - 12/22/13 at 5:59am
post #44 of 46
Quote:
Originally Posted by kn19h7 View Post
 


Not really, SE535 has serious phase-coherence problem and it can also be seen in 300Hz square wave graph. (based on graphs at innerfidelity)

 

 

red = Shure SE535

blue = Sennheiser HD800

 

The Shure SE535 probably has it's own flaws  (it's only  a $400 earphone). But the Shure SE535 is beating the HD800 at square wave tests.

 

All I'm saying is, you can't really listen with your eyes. 

post #45 of 46
Quote:
Originally Posted by ubs28 View Post
 

 

 

red = Shure SE535

blue = Sennheiser HD800

 

The Shure SE535 probably has it's own flaws  (it's only  a $400 earphone). But the Shure SE535 is beating the HD800 at square wave tests.

 

All I'm saying is, you can't really listen with your eyes. 

Can you see the downward movement before going upward (or upward movement before going downward) on the SE535's line? That's a sign of something going really out-of-phase, and that's why I said "not really" better.

 

I am not saying like you can listen with eyes, just that your example isn't very good.

 

By the way, I am more towards checking if there is any serious flaws rather than how well certain metrics measured when looking at headphones measurements.. (like for SE535, you can see the flaw clearly in phase graph)


Edited by kn19h7 - 12/22/13 at 8:13am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Audiophiles - Listening to music or to gear?