Yes, but this is you consciously deciding to add something to the replay. What Diapson is asking Tim is how he knows that he is getting an experience of the actual recording as-is without any snazz, as you say? Is his criteria for this judgement of accuracy purely measurement based & how is he ensuring this? Same questions apply to you, I guess?
Introspection and hyperbole control
- Thread starter Ron Resnick
- Start date
Yes, but this is you consciously deciding to add something to the replay. What Diapson is asking Tim is how he knows that he is getting an experience of the actual recording as-is without any snazz, as you say? Is his criteria for this judgement of accuracy purely measurement based & how is he ensuring this? Same questions apply to you, I guess?
From listening to a lot of systems, and a lot of music just like everybody else.
I take no offense to what you're saying, but I do have a point of view A) Accuracy is not subjective. But "subjective accuracy" is often what people get to when they can't admit that the like coloration. B) I will admit that I like coloration. I've had headphones that are about as close to flat as a transducer can get. Without room gain, they don't sound right, so all my favorite headphones have significant coloration.
Plenty respectful, thanks. Yes, in speakers, objective accuracy sounds good to me. As I already stated, I don't believe accuracy is subjective.
Yes, if I have low noise and distortion, and smooth, linear response on and off axis in a good-sounding room, I assume "unpleasant" is the recording. Or the music itself.
I tell them that there is no such thing as "subjective accuracy." That's like subjective objectivity. An oxymoron. And I tell them that iif they have a system that is significantly colored, it will paint every recording in their collection with the same brush. Warmth will make bright recordings sound warmer; it will make warm recordings sound dull. I tell them that equalization could be their friend, that they could have their cake and eat it too.
No offense taken, and there's my view.
Tim
Sort of. I don't think we should use enhancement, but I am asking whether it is sometimes baked into a system that uses archaic technology to produce accidental 'euphonic distortion' or extra 'warmth'. And if a system's capabilities are limited, is some form of enhancement, in fact, essential? If so, which is the more important information to a potential purchaser: that a reviewer loves the 'enhancement', or that the system was limited and needed enhancing in the first place?
In another tortured analogy I am posing two scenarios:
1. A life size print of The Mona Lisa in very high definition, colour matched to the original.
2. A print of the Mona Lisa, reduced in size, with a blue tint, and surrounded by a very ornate frame.
A reviewer finds himself incredibly moved by (2) and describes how it affects his emotions. Scenario (1) is magnificent, but with (2) he sees the painting afresh. The reduced size, colouration and the larger frame suggest a vulnerability in the sitter that he had never considered before. He suddenly understands the fragility of life itself etc. etc. etc.
Is this of any value at all? I would say not. Of far more value is the objective specs, and that with (1) the viewer gets the image as the artist created it, and can take from it what he will - if it is a 'slow burn' to learn to appreciate it, then so be it.
I am delighted that the dialogue prompted by my original post has resulted in what I think is one of the most thoughtful and intellectual discussions on WBF.
And this is what Diapson was getting at - you have no objective judgement of accuracy that you use - you do it like everyone else does & decide which sound you "think" is more accurate, which sound you prefer. You don't really know whether you are listening to a "pleasant distortion" or not.
To quote diapson "What do you say to other people who can't find subjective accuracy in equipment that measured well?"
Tim, I believe the point is that ultimately you don't "know" if you system, in your room is accurate - you don't have a set of objective measurements which fully characterises your system in your space. Therefore "accuracy" is a subjective judgement you make based on what you hear.
OK, I take a different view about this - if the enhancement sounds more realistic then I try to analyse what ingredient is missing from the original sound which is needed - why is it missing & why is it needed. This can then feed into a better capture & rendition of a more realistic sound
Yes, the analogy is tortured
If we want to go with paintings Vs reproductions of paintings & try to equate it to what is being talked about in audio here, I would suggest something along these lines:
- When viewing the original painting you can examine the brush strokes, the thickness of the paint, etc & form a view about it's method of production & emotion behind the painting. If you have ever seen later Van Gogh paintings you will see these aspects quiet easily.
- Now a 2D print of this painting loses this texture & this information
- a print which is a faux painting & has been done over a randomly textured canvas can give a feeling of more realism but it is an artifact, it doesn't match the original texture of the brushstrokes or the depth of application of the paint yet seems to be more realistic - just not realism to the original painting
- a print which has been textured as the original painting (in other words the brush strokes & paint layering are exactly copied) brings us much closer to the original & the same examination & emotional connection can be evoked.
- we don't need to analyse the original paint used & recreate the formula to use on the print in order to get the feeling of originality from the print - thi would be a waste of time & wouldn't add to the impression of realism. Similarly we don't need to analyse & recreate many aspects of the original painting in order to approach the perception of realism - we don't need the canvas/frame itself reproduced
- in other words we don't need to exactly copy everything, just the relevant elements that are important to our visual perception that convey realism.
Does this capture some semblance of the various degrees of audio realism that I'm talking about in our replay systems? The analogy breaks down as all analogies do - by the fact that a painting is a static object & audio is a dynamic piece. Furthermore we know that capturing the original brushstrokes & layering of paint are going to make the reprint more realistic - we know what is needed to be captured in order to come closer to the realism. In audio, in the absence of knowing what are the important aspects of the soundfield to our auditory perception, we try to capture everything accurately & play it back with accuracy, which is an impossible task given the systems we use.
Which is where we come to the debate so often encountered - one side claims the accuracy is enough & the other side claims they can hear problems or hear improvements with certain devices.
Last edited:
OK, substitute "laser-scanned, 3D-printed replica" for "high def print".
I don't worry too much about the dynamic vs. static issue: a painting is still dynamic in the eye of the beholder - in how they view the painting (eye sweeping over the canvas, or homing in on one detail, interpreting the philosophical meaning or examining the brush strokes).
Where I think the whole issue of 'objectivity' breaks down is that the specs that people look at are incomplete. It is far better to examine the design of the system than to measure it and attempt to interpret the readings. For example, as I mentioned before, I believe that there is a clear difference in sound between passive and active speakers, but I know of no simple specification or set of specifications that would show up that difference. Which is not to say that it cannot be measured, merely that it is a multidimensional characteristic that cannot be translated into a couple of numbers and some 2D graphs.
OK, I was using old-school techniques, I guess
But in your analogy you state "Of far more value is the objective specs" & yet below you state that the "specs are incomplete". So what are we to conclude? You "think" the laser-scanned image is accurate but current measurements/specs do not tell you this & therefore you have no way of knowing this?
Your view about static Vs dynamic is very wrong, IMO but what you say does illustrate one aspect of perception - attention & focus. We have different ways of perceiving - a holistic view where we stand back & try to "take in" the painting as a whole, use of light & shade, examine our gut reaction to it & emotional aspects of it that might hit us Vs the much more focussed analysis on a particular aspect of the painting, brush strokes, painting technique, layering, etc. & we are oblivious, in that moment, to other aspects. Yes we can quickly flit between these different attention modes but this is made easier because it is a static object we are viewing. Perceptions work in the same way in audio except that we are dealing with a dynamic object which makes it much more complicated & taxing - it decreases our working memory.
Agreed about the specs - they are inadequate to determine how something sounds - but not sure if examining the design would prove profitable, in this context & I will leave the difference between active & passive speakers to others - I imagine many will argue that the difference between these is measureable
No, the point is that Diapson asked some specific questions, and I gave him specific answers. I believe electronics should be as true to the recording as possible, within budget. They used to call that high-fidelity. I believe it's a lot harder with speakers, but they can be pretty good if the designers know what they're doing and make smooth, linear response their primary goal. And I believe in good-sounding rooms. Now, in the last step, "accuracy" suffers. Enjoyment doesn't need to. And like I said in my post to Diapson, I'm ok with that. My ears are used to hearing in rooms. My perception likes a certain amount of room gain. And that part is admittedly subjective. But I don't personally believe starting out with "colored" media or electronics, because, as I said above, it colors everything with the same brush. How do I know when electronics are not colored? Well, we both know the answer to that, but we'd have to take it over to the Science of Audio forum.
Tim
Hi Tim,
I think like jkeny, I’m still finding this slightly problematic. Your system - “low (in) noise and distortion, and (with) smooth, linear response on and off axis in a good-sounding room” is only ever theoretical. It doesn’t exist in the real world because it denies these two levels of reality:
1) the steady-state signals used to produce measurements “low (in) noise and distortion and smooth, linear response on and off axis” are not the signals of a musical, dynamic and harmonically complex waveform constantly modulating over time;
and,
2) the system itself is electronically (and in the case of all transducers - electro-mechanically) interdependent: that is, it is interacting with itself over time as, for instance, the amplifier’s current modulates the speaker and the speaker’s impedance modulates the amp, the stylus modulates the arm and the arm modulates the stylus. Our components are never not working independently isolated of one another (as they are on the test bench - the “known knowns”), but are always highly dependent on one another - the mechanism cannot function when broken down to its constituent parts, each component must be placed into an electro-mechanical relationship with the another, the result of which (the “known unknowns”) is highly volatile in nature and will therefore produce measurements well outside those originally produced in the above scenario.
It’s one thing to say the individual components of our system are “accurate” within the context of their objectively measurable audible limits when playing steady-state signals through them, but it’s a wholly different thing to state with any degree of confidence that our systems can be accurate to the recording unless we’re willing to directly compare the musical waveform of the recording medium against the same musical waveform when captured at the speaker/room interface.
Last edited:
So all things are variable, and we cannot know anything? Designers can't understand the listening-room impact of the measurable differences between, say, vinyl and a digital file of the same master? What we can hear cannot be measured in any way that is meaningful relative to listening? Bummer....nah, it's ok. I have a quiet afternoon ahead of me. I'll just listen to some good music.
Tim
You miss the point, Tim - system thinking & relevant measurement techniques are required is the point. Without this your "faith" in the accuracy of your playback system is as subjective as anyone else.
We can certainly objectively “know” how a component measures on a bench, in isolation, when tested with steady-state signals, yes.
Had you said, “I believe electronics should be as true to the test signals as possible…” then we need only look at the measurements to establish the degree of objective accuracy.
But if we persist in asserting that electronics should be as true to the recording as possible, then we need to use the recording to determine the veracity against which those measurements can be considered meaningful (not steady-state signals), and we cannot do that unless they are inserted into a system of interdependence whereby the medium and the mechanism are inextricably linked and dynamically interacting with one another.
Again, if the latter rather than the former is what you believe, then simple: we take the waveform of the medium (the recording) and compare it to the waveform of the mechanism (the system) at the speaker/room interface.
Well, here's a sine wave typically used on the test bench:
And here's Frankie Goes to Hollywood's "Welcome to the Pleasuredome - Fruitness" taken from their 12 XZTAS 7 EP (I'm on a bit of Trevor Horn kick right now):
Which of the two above would you say would produce a more-linear result?
Last edited:
Yes, I can understand your view about specs/measurements but then you should understand my & others view about the test signals being used to derive these measurements are not representative of the final signals that you will use with the system.
This has particular relevance because the devices in the system & their interactions are non-linear - therefore measurements with linear signals cannot be extrapolated to predict this non-linear behaviour.
Love to discuss with you on the science thread if you so choose, I think we see things overall the same just the way to get there differently. ie, sweeped sinewaves and imd is more a comparison to what you show.
Honestly, I don't know that I'd have much more to contribute than what I have here already.
Perhaps swept sine waves are more relevant, but even then they don't contain anywhere near the harmonic complexity of music, which in a stereo signal can vary greatly between the two channels in overall frequency and overall note value. The FGTH track above, as an example, at 4:10.217, has an overall frequency of 100.76 Hz and an overall musical note value of G2 +48 cents in the left channel, while the right channel at the same point in time has an overall frequency of 339.19 Hz and an overall musical note value of E4 +49 cents. This is the spectral pitch display:
Contrast this with the sine wave I posted above which has an overall frequency of 446.02 Hz and an overall musical note value of A4 +23 cents:
Even this small slice of time doesn't take into account the amplitude and time modulations taking place simultaneously. Music is all three of these variables (pitch, time, amplitude) in constant modulation, a swept sine wave contains only one of these modulations (pitch). Of course, I could have chosen music with a less complex arrangement - a solo kazoo for instance - but in the real world, people use their hi-fi to play all sorts of stuff, not necessarily limited to over-produced, indulgent, 80's anthems to debauchery and the poetry of Samuel Taylor Coleridge.
You said, in the post above, that I had missed the point, but now you seem to be making it again. You appear to be saying that there is not a meaningful relationship between the measurements engineers use to asses, develop and refine their products, and the impact that the data has on a component's performance when playing a recording through a system. Did I get that right?
Tim
Similar threads
- Locked