Is transient response the most important thing for the perceived audio quality in a system ?

[Cbdb2]

@RayDunzl
take that snare attack and repeat it every milisec, it won't sound like a snare.

 

[Cbdb2]

on its own, it resembles a softly played key,
within the context, it gives the impression of a synth due to the noticeable absence of soft playing.
However, that aspect is not particularly important, as the key point remains that the original character has been altered.

You said
"It seems that the very first sound ( transient ) from a flute or acoustic guitar makes the caracter of the instruments timbre and also maybe the perceived pitch ?

These are both incorrect as I was trying to show above.

 

[Axo1989]

Very true. I think your harpsicord example is a good illustration of what your speakers and electronics are capable of vs. how you hear... Fourier says that playing the track forwards vs. backwards makes no difference to the speakers and electronics. Both forward and backward play have exactly the same 'transients', i.e. Fourier components! There are no 'attacks' or 'decays' that exist playing reverse vs. playing forward. Your speakers don't care either way. The transients area all there but the ear and brain hear much different sounds; that's the psychoacoustics part!!!

Aren't you talking past each other somewhat?

The ADSR envelope (or whatever extended version of the acronym you prefer) of the harpsichord (or another instrument) isn't merely psycho-acoustic (or it isn't psycho, it is acoustic). Playing in reverse will reproduce a different (hopefully mirrored) envelope and waveform. We don't just imagine it sounds different, it does.

I think confusion about transients also occurs between two aspects, rise time and amplitude. I'd say the attack can't (by definition) be faster than the rise time of the highest frequency component. The period of a 20 kHz wave (in air at 20º C) is 0.00005 seconds or 50 microseconds, so 50/4 or 12.5 µs for the initial rise (correct me if that's wrong). It's good to have something on the back of the envelope (forgive the pun) in order to speculate.

The theoretical appeal of an extended high frequency tweeter is that it may achieve the faster rise time corresponding to say 40 kHz (the specification for mine, for example). So do acoustic or electronic instruments need that? We do know that acoustic instruments can produce ultrasonics, so maybe, and electronic instruments can theoretically do anything. And if they do can we hear it? Apparently we can perceive inter-aural time differences of 20 µs* (assuming some equivalence between this and perception of transients) so we probably fall short (if that is the lower limit and my preceding maths is ok) but not by as much as I thought.

The other aspect of the transient is amplitude. Can the speaker reproduce the dynamic peak of the instrument? For some instruments, certainly not all speakers can (let's ignore amplification to keep it simple, but there will be limits there as well). There are arguments that we shouldn't listen to that, but people also misunderstand that amplitude plus duration are required for hearing damage, not amplitude alone (that said, drummers wear earplugs, there's the matter of proximity).

So my lay interpretation is that timing is likely resolved but dynamics often isn't. Now I'll post this and wait for my beating/re-education.

*per Blauert J (1997) Spatial Hearing: The Psychophysics of Human Sound Localization. Cambridge, MA: MIT Press (which I've seen cited but not read) for those of you that require citations.

 

[dasdoing]

You said
"It seems that the very first sound ( transient ) from a flute or acoustic guitar makes the caracter of the instruments timbre and also maybe the perceived pitch ?

No I didn't

 

[dlaloum]

I think a lot of what people call transiet response is really peak SPL capacity. Any system that can reproduce 20 KHz is "fast" enough for any transient but you also need to be able to have enough SPL headroom to play the music peaks without compression or distortion which depending on the music can be quite demanding.

I agree that a lot of people conflate dynamic range & peak SPL, with transient response....

But have a listen to something like a set of Quad ESL63's (or the current ESL2812) - these are wonderful demonstrations of the importance of micr0-detail, and a lot of that comes down to transient response...

A well-nigh weightless electrostatic driver, results in very very good transient response natively.

This then means you get a level of detail and micro-detail in the sound which (to my ears) makes it very real - very "high fidelity" - you can record a live acoustic event, play it back on this type of speaker, and the illusion of reality is very authentic indeed.

This in no way relates to dynamic peak capabilities - most electrostatics are severely limited in terms of peak output... (some attempt to compensate by using very large or multiple panels... but basically electrostatic drivers have very limited motion, hence very limited peak spl)

There are other technologies that also work towards the ideal of a weightless driver... ribbons, etc...

Very few of these are capable in the bass regions, with even very large ESL's being bass constrained.

However, within those constraints, the bass produced by pure ESL's tends to be very pure - very low distortion... and excellent in transient terms.

Instruments with loads of low frequency harmonics (eg: double bass) - sound real, on a set of Quad ESL's in a way that most standard "dynamic" (ie: cones & magnets) speakers struggle to achieve.

When my main speakers were ESL's I struggled to find a Subwoofer that would sound "right" with the ESL's and make it sound like a single speaker, rather than a speaker with a sub tacked on... I have long theorised that the issue was probably the relatively high level of distortion in most subs, and their inability to reproduce transients the way an ESL does... ie: they are too damn slow!

I did find that the subs that got closest to a match with the ESL's were sealed types... and also had relatively low distortion stats as well.

No there were no measurements.... and I last tried this more than 15 years ago, so it should be taken with whatever sized pinch of salt the reader chooses.

However, in high fidelity terms, my experience with ESL's indicates (to me) that transient response is critical to accurate audio reproduction.

With artificial environments / recordings... including both music and movies, this is less of an issue, as most of what is heard there is inherently artificial... our ears and mind have no reference point.

But play a solo acoustic instrument with which the listener is familiar, and transient response becomes one of the key things which seperates good speakers from GREAT speakers.

 

[IPunchCholla]

With artificial environments / recordings... including both music and movies, this is less of an issue, as most of what is heard there is inherently artificial... our ears and mind have no reference point.

Why is this true? What mechanism of our brain allows it to use transients (which I’m not all convinced we’re talking about the same thing) to be able to hear acoustic instruments better? Or are you arguing the opposite? What does the source of the recording being acoustic have to do with the speaker?

Let’s say we have 2 identical speakers. Identical in every way except the ability to produce transients. As we know, the ability to do this is totally the ability to produce frequency. So if transients steeper than 20khz matter, then that speaker would also have a different FR. In particular the ability of the speaker with “better” transients would have to be flatter out some significant number if kHz. So this should be easy-ish to test.

 

[HarmonicTHD]

Gentlemen:

Here are four definitions of "transient" in acoustics from the web. Are they correct?

Transient (acoustics) - Wikipedia

en.wikipedia.org

What is a Transient in Audio Production?

Learn about what a transient is, what it sounds like, and how to optimize transients in your mix to get a professional sound.

www.izotope.com

What Are Transients in Music & Why Are They Important? (Explained)

In the world of audio engineering & music production, there are a lot of terminologies you need to understand. While your ears are the most important

www.whippedcreamsounds.com

Transient - inSync

A non-repeating waveform, usually of much higher level than the surrounding sounds or average level. Good... Read more »

www.sweetwater.com

Jim

Yes.

 

[Blumlein 88]

I agree that a lot of people conflate dynamic range & peak SPL, with transient response....

But have a listen to something like a set of Quad ESL63's (or the current ESL2812) - these are wonderful demonstrations of the importance of micr0-detail, and a lot of that comes down to transient response...

A well-nigh weightless electrostatic driver, results in very very good transient response natively.

This then means you get a level of detail and micro-detail in the sound which (to my ears) makes it very real - very "high fidelity" - you can record a live acoustic event, play it back on this type of speaker, and the illusion of reality is very authentic indeed.

This in no way relates to dynamic peak capabilities - most electrostatics are severely limited in terms of peak output... (some attempt to compensate by using very large or multiple panels... but basically electrostatic drivers have very limited motion, hence very limited peak spl)

There are other technologies that also work towards the ideal of a weightless driver... ribbons, etc...

Very few of these are capable in the bass regions, with even very large ESL's being bass constrained.

However, within those constraints, the bass produced by pure ESL's tends to be very pure - very low distortion... and excellent in transient terms.

Instruments with loads of low frequency harmonics (eg: double bass) - sound real, on a set of Quad ESL's in a way that most standard "dynamic" (ie: cones & magnets) speakers struggle to achieve.

When my main speakers were ESL's I struggled to find a Subwoofer that would sound "right" with the ESL's and make it sound like a single speaker, rather than a speaker with a sub tacked on... I have long theorised that the issue was probably the relatively high level of distortion in most subs, and their inability to reproduce transients the way an ESL does... ie: they are too damn slow!

I did find that the subs that got closest to a match with the ESL's were sealed types... and also had relatively low distortion stats as well.

No there were no measurements.... and I last tried this more than 15 years ago, so it should be taken with whatever sized pinch of salt the reader chooses.

However, in high fidelity terms, my experience with ESL's indicates (to me) that transient response is critical to accurate audio reproduction.

With artificial environments / recordings... including both music and movies, this is less of an issue, as most of what is heard there is inherently artificial... our ears and mind have no reference point.

But play a solo acoustic instrument with which the listener is familiar, and transient response becomes one of the key things which seperates good speakers from GREAT speakers.

I've had the same experience. It does seem fast. An example is hearing a finger plucked acoustic guitar. Notes are not that loud nor is the frequency very high. Hearing the real thing versus very good cone and box speakers it sounds like nearly the 1st half of that sound was over before the speaker caught up to it. Poor box speakers are worse. Hearing this over an ESL 63 there is just the barest loss like you got 7/8ths of the sound just barely less than reality. I've tried this with a good microphone directly into speakers with the acoustic guitar there for the comparison directly. Other speakers sound sluggish with this fairly simple sound.

I think the ESL 63 does this somewhat better than other ESL's I've had. Better than 57s, better than Acoustats and better than Soundlabs. All of those seem faster to respond than box speakers.

 

[dualazmak]

When my main speakers were ESL's I struggled to find a Subwoofer that would sound "right" with the ESL's and make it sound like a single speaker, rather than a speaker with a sub tacked on... I have long theorised that the issue was probably the relatively high level of distortion in most subs, and their inability to reproduce transients the way an ESL does... ie: they are too damn slow!

I essentially agree with you.

Even after we would have decided which sub-woofer (SW) and woofer (WO) to be used in our audio setup, I believe it is really important and critical to establish time alignment between the two, at most suitable XO Fq, with most suitable XO slopes; low-pass (high-cut) for SW and high-pass (low-cut) for WO.

Even having rather steep slope filters, we always have Fq window where SW and WO would audibly sing together, and therefore the tuning of transient behaviors around that Fq window is critical for excellent total transient response of your audio system.

We can semi-objectively observe by our eyes using Adobe Audition's 3D (time-gain-Fq) spectrum of the recorded sound for such tuning procedures for;
1. XO Fq selection,
2. XO filter type and slope selections,
3. Gain matching/adjustment between SW and WO,
4. Time alignment adjustment in msec precision.

As for precision time alignment between SW and WO, I utilized rather primitive but validated and reliable "time-shifted tone burst measurement method" as shared in my post here.

For measurement and tuning of XO configuration for SW and WO, together with transient behavior observation/confirmation, Adobe Audition's 3D (time-gain-Fq) spectrum has been a powerful tool in my case, as shared in my project thread;
- Measurement of transient characteristics of Yamaha 30 cm woofer JA-3058 in sealed cabinet and Yamaha active sub-woofer YST-SW1000: #495, #497, #503, #507

Since I know that single photo or diagram is worth 1,000 or more words, let me share here only two diagrams selected from my project thread on this issue of SW-WO best matching for excellent total transient response;

Having these and other intensive objective measurements and of course also intensive subjective hearing sessions, in my setup with SW Yamaha YST-SW1000 and WO Yamaha 30 cm woofer JA-3058 in NS-1000's sealed cabinet, I decided the XO Fq (50 Hz) and the filters as summarized here and here on my project thread.

I believe this is also true for other SP drivers, WO-to-Midrange-to-Tweeter-to-SuperTweeter, as I have intensively done in my multichannel project. Please refer to my summary post here here and here.

In any way, various tuning towards better "transient behavior/response" of our audio setup does "matter" and is important for the "perceived audio quality in a system" (the wording of the thread title here).

 

[dlaloum]

Why is this true? What mechanism of our brain allows it to use transients (which I’m not all convinced we’re talking about the same thing) to be able to hear acoustic instruments better? Or are you arguing the opposite? What does the source of the recording being acoustic have to do with the speaker?

Let’s say we have 2 identical speakers. Identical in every way except the ability to produce transients. As we know, the ability to do this is totally the ability to produce frequency. So if transients steeper than 20khz matter, then that speaker would also have a different FR. In particular the ability of the speaker with “better” transients would have to be flatter out some significant number if kHz. So this should be easy-ish to test.

It has nothing to do with the speakers - it has everything to do with our ears as a measurement instrument - we have auditory memory of instruments we are familiar with - it is a benchmark - many of us have the instruments at home, and can directly play/compare.... or alternatively use something simple and extremely familiar - such as your wife's voice.

There is a direct, and real reference - you can hear whether it sounds "real" or not. And even to some degree, "how" real it sounds.

With synthesised sounds, there is no reference, there is no benchmark, there is nothing for your ears/brain to latch onto and compare.

You can say it sounds good or bad, I like it or dislike it, but you cannot say - it sounds like X - because X is completely synthesised, you therefore have no auditory memory benchmark.

You may be able to measure aspects of the synthesised (and acoustic recorded) sounds, as well as the speaker performance, but whether that in fact works in terms of our psycho-acoustic response, is very much an open question... we still don't have the details of what is required to properly deceive our psycho-acoustic system.

One thing I can say from experience (subjective, yes!) - is that certain speakers (electrostatics being an obvious example genre!) - can do a very very good job of reproducing auditory reality, and to the level of tricking people as to whether things are actually in the room (or in the next room)....
Most speakers don't achieve this illusion. Most speakers when playing a recording of my partners voice, won't convince me that it is in fact her.

There are multiple variables at play, that is certain - but all the pure electrostatic speakers I have listened to, (ie without a "dynamic" woofer, not a hybrid design) - have had the ability to pull off that illusion.

This results in certain sounds being reproduced with uncanny realism, things like doors opening or closing, footsteps, wind in the trees, leaves rustling.... - these are the places where transient response (I am making some assumptions that it is transient response that is the key factor) - becomes an identifier, a marker, of speakers that can pull off the auditory illusions that mark a really top high fidelity system.

And it is worthy of note, that in terms of frequency extension (both top and bottom) some of these ESL designs are constrained... yet their transient response remains excellent.... and they are also a design type that is capable of passing a square wave - a feat that is in almost all cases out of reach of most speakers.

I'm not going to get engaged in the deep mathematics of transient response, phase, frequency and fourier transforms.... these are interesting, and allow analysis of aspects of design and performance - but ultimately don't tell you much about the psycho-acoustic response, the ability of the speaker to pull off the illusion of reality. (which is of course only valid in cases where the signal being played back originates in reality... )

 

[IPunchCholla]

Gentlemen:

Here are four definitions of "transient" in acoustics from the web. Are they correct?

Transient (acoustics) - Wikipedia

en.wikipedia.org

What is a Transient in Audio Production?

Learn about what a transient is, what it sounds like, and how to optimize transients in your mix to get a professional sound.

www.izotope.com

What Are Transients in Music & Why Are They Important? (Explained)

In the world of audio engineering & music production, there are a lot of terminologies you need to understand. While your ears are the most important

www.whippedcreamsounds.com

Transient - inSync

A non-repeating waveform, usually of much higher level than the surrounding sounds or average level. Good... Read more »

www.sweetwater.com

Jim

That is my understanding. So the only question that seems relevant, given my understanding of Nyquist is how much distortion free bandwidth does a speaker need in order to produce the transients we can hear?

Is there any research that indicates we need more than 44.1? I’m aware of one study using ultrasonics that people could feel a difference, but there seems to be some disagreement. I’m also aware of the experiments that show we can’t here a missing fundamental if the harmonics that create it are beyond our ability to hear.

On the flip side, can’t nyquist be reversed? If I can produce an infinite (bandwidth unlimited) number of sine waves, I can produce any waveform. So at best isn’t the division between acoustic and electronic just execution of technology and not an inherent divide?

 

[IPunchCholla]

It has nothing to do with the speakers - it has everything to do with our ears as a measurement instrument - we have auditory memory of instruments we are familiar with - it is a benchmark - many of us have the instruments at home, and can directly play/compare.... or alternatively use something simple and extremely familiar - such as your wife's voice.

There is a direct, and real reference - you can hear whether it sounds "real" or not. And even to some degree, "how" real it sounds.

With synthesised sounds, there is no reference, there is no benchmark, there is nothing for your ears/brain to latch onto and compare.

You can say it sounds good or bad, I like it or dislike it, but you cannot say - it sounds like X - because X is completely synthesised, you therefore have no auditory memory benchmark.

You may be able to measure aspects of the synthesised (and acoustic recorded) sounds, as well as the speaker performance, but whether that in fact works in terms of our psycho-acoustic response, is very much an open question... we still don't have the details of what is required to properly deceive our psycho-acoustic system.

One thing I can say from experience (subjective, yes!) - is that certain speakers (electrostatics being an obvious example genre!) - can do a very very good job of reproducing auditory reality, and to the level of tricking people as to whether things are actually in the room (or in the next room)....
Most speakers don't achieve this illusion. Most speakers when playing a recording of my partners voice, won't convince me that it is in fact her.

There are multiple variables at play, that is certain - but all the pure electrostatic speakers I have listened to, (ie without a "dynamic" woofer, not a hybrid design) - have had the ability to pull off that illusion.

This results in certain sounds being reproduced with uncanny realism, things like doors opening or closing, footsteps, wind in the trees, leaves rustling.... - these are the places where transient response (I am making some assumptions that it is transient response that is the key factor) - becomes an identifier, a marker, of speakers that can pull off the auditory illusions that mark a really top high fidelity system.

And it is worthy of note, that in terms of frequency extension (both top and bottom) some of these ESL designs are constrained... yet their transient response remains excellent.... and they are also a design type that is capable of passing a square wave - a feat that is in almost all cases out of reach of most speakers.

I'm not going to get engaged in the deep mathematics of transient response, phase, frequency and fourier transforms.... these are interesting, and allow analysis of aspects of design and performance - but ultimately don't tell you much about the psycho-acoustic response, the ability of the speaker to pull off the illusion of reality. (which is of course only valid in cases where the signal being played back originates in reality... )

Sure. But my readings on acoustic memory seem to indicate that while acoustic memory for repeated events is accurate, it is also imprecise. This means that while we are very good at sounds like a trumpet or your mom, it still sounds like a trumpet or your mom, even if we are hearing are mom from before we are born or in the nursing home. So long term acoustic memory is very fallible although a powerful force.

I can create a waveform electronically and listen to it over and over on different speakers and develop that same long term memory of being able to recognize it, even though the form is being changed uniquely by every transducer. Reference to the acoustic isn’t required.

 

[Axo1989]

Sure. But my readings on acoustic memory seem to indicate that while acoustic memory for repeated events is accurate, it is also imprecise. This means that while we are very good at sounds like a trumpet or your mom, it still sounds like a trumpet or your mom, even if we are hearing are mom from before we are born or in the nursing home. So long term acoustic memory is very fallible although a powerful force.

I can create a waveform electronically and listen to it over and over on different speakers and develop that same long term memory of being able to recognize it, even though the form is being changed uniquely by every transducer. Reference to the acoustic isn’t required.

I was going to say something similar. We can presumably form memory of any sound, some people recognise a particular synth for example and differentiate it from others. So we had original Roland TR808 and their own recent clone TR08 reproducing its timbres.

 

[dlaloum]

Sure. But my readings on acoustic memory seem to indicate that while acoustic memory for repeated events is accurate, it is also imprecise. This means that while we are very good at sounds like a trumpet or your mom, it still sounds like a trumpet or your mom, even if we are hearing are mom from before we are born or in the nursing home. So long term acoustic memory is very fallible although a powerful force.

I can create a waveform electronically and listen to it over and over on different speakers and develop that same long term memory of being able to recognize it, even though the form is being changed uniquely by every transducer. Reference to the acoustic isn’t required.

In the case of the example you are using - the electronic waveform is your own benchmark, you have developed it, and know how it sounds - you will have an auditory benchmark - and can use it on other speakers.

But if someone else has developed an electronic waveform (ie: any form of synth sound... so it's not exactly an edge case!) - you have no idea as to how the composer/author expected it to sound. - He originally heard it through his own headphones or speakers and room, affected by the flaws of that environment... The true benchmark - was/is NOT the waveform - but the acoustic experience that the author had and approved of as composer.

That is why is do suggest to people that they record the voice of someone they know intimately (like your mom)... and use that as a benchmark - for all the flaws of psycho-acoustic memory... we can and do identify voices with great accuracy.... and we can do various experiments on that voice recording, to distort transients, frequency response, phase, or other aspects, and see whether it still sounds "real".

The test is the illusion, not the graph/measurement.... but if you can find the measurement which differentiates "illusion" from "failed illusion" - then you have something measurable and repeatable which can then be used to progress speakers/dsp/room design and setup.

 

[IPunchCholla]

In the case of the example you are using - the electronic waveform is your own benchmark, you have developed it, and know how it sounds - you will have an auditory benchmark - and can use it on other speakers.

But if someone else has developed an electronic waveform (ie: any form of synth sound... so it's not exactly an edge case!) - you have no idea as to how the composer/author expected it to sound. - He originally heard it through his own headphones or speakers and room, affected by the flaws of that environment... The true benchmark - was/is NOT the waveform - but the acoustic experience that the author had and approved of as composer.

That is why is do suggest to people that they record the voice of someone they know intimately (like your mom)... and use that as a benchmark - for all the flaws of psycho-acoustic memory... we can and do identify voices with great accuracy.... and we can do various experiments on that voice recording, to distort transients, frequency response, phase, or other aspects, and see whether it still sounds "real".

The test is the illusion, not the graph/measurement.... but if you can find the measurement which differentiates "illusion" from "failed illusion" - then you have something measurable and repeatable which can then be used to progress speakers/dsp/room design and setup.

Ok. I think I was reading an acoustic vs electronic relationship to transients when there wasn’t one.

Just as a reference point, I played trumpet for a few years. I have a very strong memory for what that sounded like. I have heard Kermit Ruffins play in a small club. I could not say whether his trumpet sounded “real” as compared to my ingrained memory of a trumpet. It sounded different. Neither sound playback of trumpet on my speakers, but that is because I’m not listening loud enough, most of the time. When I turn it up, most differences disappear.

So I think level matching to the “real” thing would be an important first step, which then indicates distortion free (below 1%) and FR level to 20khz at 95 dBSPL @ 3 meters is the goal. If we can do that then perhaps looking at increasing the bandwidth to get “better” transients would be a good thing to explore.

 

[dualazmak]

That is why is do suggest to people that they record the voice of someone they know intimately (like your mom)... and use that as a benchmark - for all the flaws of psycho-acoustic memory... we can and do identify voices with great accuracy.... and we can do various experiments on that voice recording, to distort transients, frequency response, phase, or other aspects, and see whether it still sounds "real".

The test is the illusion, not the graph/measurement.... but if you can find the measurement which differentiates "illusion" from "failed illusion" - then you have something measurable and repeatable which can then be used to progress speakers/dsp/room design and setup.

Yes, I fully agree with you.

I too believe subjective hearing assessment of our audio system (including room acoustics) would be always critical basic fundamentals, and as you kindly pointed, we need to have our own consistent and seldom-to-be-changed "audio sampler/reference music/sound playlist" throughout our audio exploration journey.

Your suggestion of "mom's voice" is really wonderful example, I believe!

I believe, furthermore, our consistent "audio sampler/reference music/sound playlist" should consist of several tracks, at least 10 tracks I assume (including mom's voice!), suitable for check and assess many factors in the "perceived audio quality of a system"; the playlist may be selected also based on your music preferences.

Just for your reference, my present multi-purpose "audio sampler/reference playlist" consists of 60 tracks as summarized and shared in my posts here and here on my project thread.

 

[dlaloum]

Ok. I think I was reading an acoustic vs electronic relationship to transients when there wasn’t one.

Just as a reference point, I played trumpet for a few years. I have a very strong memory for what that sounded like. I have heard Kermit Ruffins play in a small club. I could not say whether his trumpet sounded “real” as compared to my ingrained memory of a trumpet. It sounded different. Neither sound playback of trumpet on my speakers, but that is because I’m not listening loud enough, most of the time. When I turn it up, most differences disappear.

So I think level matching to the “real” thing would be an important first step, which then indicates distortion free (below 1%) and FR level to 20khz at 95 dBSPL @ 3 meters is the goal. If we can do that then perhaps looking at increasing the bandwidth to get “better” transients would be a good thing to explore.

My target is exactly that - 95db max SPL, my normal listening levels (average volume) is around 75db, and allowing for 20db peaks takes me to 95db.... but I like to provide headroom beyond that (up to circa 105db) .... just because some music has that sort of dynamics - even when they are brief!

With regards to distance from speakers - I would suggest that this varies, and depends on room setup - I listen at 2.4m.

But one of the things that years of listening to / owning electrostatics taught me, is that transients are a micro as well as a macro phenomenon - that instantaneous no lag response is as much part of a bow on a violin string at its softest, as it is that of a trumpet at full blast (which for authenticity, would require more than 95db! - thems get seriously loud!).

And the really noticeable thing that will cause surprise for a multitude of listeners, is the micro details when it is quiet.... the mom's voice test, the soft background sounds in a movie filmed in a natural environment; shoes on autumn leaves, wind in the trees, the stuff people don't normally notice, which becomes noticeable simply because all of a sudden it sounds real... suspension of disbelief is no longer required... it sounds real.

I love the fireworks, and the loud explosions too.... most halfway decent systems can do those impressively.... show me a system that can do the soft transients right, and you are showing me true high end.

(yes I really did at one point run an all electrostatic surround setup - it lasted only 4 or 5 years before WAF issues pushed it out.... sound quality was stunning.)

 

[danadam]

on its own, it resembles a softly played key,
within the context, it gives the impression of a synth due to the noticeable absence of soft playing.
However, that aspect is not particularly important, as the key point remains that the original character has been altered.

They're not volume matched, so no surprise there:

Track gain, Track peak
   1.78 dB,   0.994873, piano_C_major.wav
   2.30 dB,   0.713252, piano_C_major_-3dB.wav

What if you reduce the original by 0.5 dB (or amplify the compressed version by 0.5 dB)?

 

[dasdoing]

They're not volume matched, so no surprise there:

Track gain, Track peak
   1.78 dB,   0.994873, piano_C_major.wav
   2.30 dB,   0.713252, piano_C_major_-3dB.wav

What if you reduce the original by 0.5 dB (or amplify the compressed version by 0.5 dB)?

I'm confused. I decreased the peak by 3dB, the outcome is precisely what was anticipated. This isn't merely an exercise of "can you hear a difference." In fact, I anticipate that the difference will become even more noticeable if you increase the level of the peak level reduced file.

 

[dualazmak]

And the really noticeable thing that will cause surprise for a multitude of listeners, is the micro details when it is quiet.... the mom's voice test, the soft background sounds in a movie filmed in a natural environment; shoes on autumn leaves, wind in the trees, the stuff people don't normally notice, which becomes noticeable simply because all of a sudden it sounds real... suspension of disbelief is no longer required... it sounds real.

Thank you, this is one of the very important aspects/targets of our audio setup and tuning, and the "nice system transient behavior" (together with excellent S/N) would be one of the key factors even in this kind of low-gain quiet situations.

One of my system tuning targets is exactly the same as yours; aiming towards precise and clear reproduction of very delicate expressions and nuances of the recorded sounds, even though I do not use ESL units.

Just for example,,,
Reproduction/playback of extremely delicate "nuances" of female vocal "pronunciations";

Spoiler

I wrote here;
Her (Paula Morelenbaoum's) face and lips should be always firmly allocated in slightly-upper in the center of the stereo image throughout her voice with wide frequency range and with any of pronunciations. I also check (and tune) whether I can (as if) hear how wet/dry her lips are, especially listening to her voice/words including "p", such as "As Praias Dersetas" and "Esprando por nos dois".

Very faint/subtle comfortable intrusion of environmental sounds at the recording venue such as weak wind breeze and/or extremely faint lovely singing birds;

Spoiler

I wrote here;
Furthermore, on your audio system, can you nicely hear and identify the charming birds outside of Jobim's house joining at 2:40 in this beautiful track? Even though I cannot remember where he wrote or spoke, Ryuichi Sakamoto once commented with his deep emotion that "When I carefully listen to Jobim's music, I always feel as if various charming birds are singing in his music. It was really our miracle experience, therefore, that the birds outside of Jobim's house wonderfully joined our recording of As Praias Desertas with Mrelenbuam2."

and I wrote here for the very quiet lute duet recording;
By the way, in this track, can you clearly hear a bird, Nightingale (Luscinia megarhynchos), wonderfully joining in this midnight recording session at a small quiet chapel? It would be also a challenge to our audio system.

After our such system optimization, however, the real/actual noises in the recorded sounds (which even the recording/mixing engineers did not notice), such as low gain air conditioning noises, occasionally become uncomfortable issue in our best-tuned audio setup. Very fortunately, nowadays we have several counter measures in digital domain for such issues.

Spoiler

Please refer to my post here;
- Excellent Recording Quality Music Albums/Tracks for Subjective (and Possibly Objective) Test/Check/Tuning of Multichannel Multi-Driver Multi-Amplifier Time-Aligned Active Stereo Audio System and Room Acoustics; at least a Portion and/or One Track being Analyzed by Color Spectrum of Adobe Audition in Common Parameters: [Part-17] Excellent Quality Music Tracks, But Containing Unacceptably High Gain Low-Frequency Air Conditioning Noises; What Counter Measures Can We Have? #658

BTW, my listening position is about 3.3 m away from the surface of the SP system of my 5-way 10-CH multichannel multi-SP-driver multi-amplifier fully active stereo audio setup.

Spoiler

Please refer to here and here for the details of my latest system setup.