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

[Galliardist]

Very often, with the right setup. At least I think it has to do with transients, at least in part. It’s a startling presence of a sound source in the air, in the room in front of you. Startling is the right word, because it can really startle you for its realism. This happens mostly with small ensembles, single voices, etc., not so much with large orchestras. You can easily make out groups of instruments in an orchestra, but you don’t feel like there’s a violin there somewhere. I suppose the multiple instruments blur each other’s transients and make a more homogenous sound.

It can also happen with non-acoustic (i.e. synthesized) music. I listened to a track once (I think it was Yello), and constantly looked over my shoulder because I thought there was someone at the door. The door is on left, somewhat behind the listening chair. Sounds from behind don’t usually do that. They can be interesting effects but you can tell they’re part of the programme. But this effect was so well-crafted that it fooled me multiple times in a row. I even turned off the music to see whether it stops.

These things only come out when my system is at its peak (speakers in the right position, room treated, DSP spot on). I often experiment with different speaker positions, room treatments, Dirac filters, etc. and it is very easy to ruin this ability.

It‘s a bit like the magic that suddenly happens when you bring a telescope into perfect collimation

We're talking speakers here though. The speakers' particular ability doesn't change because the setup of the rest of the system isn't perfect. And I can assure you that low cost speakers from the 1980s can achieve that sort of effect on occasion.

I'm going to stay on the very controversial edge of this though. When I have heard some of these startling effects and oddities, they aren't "real", and I can be pretty sure that they are mostly not part of what was intended in the recording. Singers are supposed to be on stage, drummers normally at the back, guitars to the sides, and so on. First violins to the left, second to the right, cellos to left of centre, (maybe in front and slightly to the left or well to the left according to the orchestra's practice), and so on. When an instrument suddenly jumps to sound "realistic" and way off to the side, or the bass player "sits in my lap" (yes, I've had that appear to happen) I regard it as wrong. I tend to think of two channel stereo (not so surround, if that is what you have) as being a bit like a slightly 3D TV image. Things should be in place and I should be able to understand everything in context. Even if you have a strong central image for your vocalist, if you have a choir or operatic chorus involved, or different singers placed across the soundstage, that one sticking out becomes, well, a problem.

Special effects aside - the Yello example you give may be the genuine article.

Dont agree at all.
Bad loudspeakers make speech very indistinct, and robs the music from every sence of realism making the experience to listen to music very boring. Havent you heard a really good hifisystem ? The sound is very far from any bluetooth kitchen device.

My Eneby 20 loudspeaker in the kitchen sounds very dull compared to my Genelecs.

In the last few weeks I have listened to music:

• on hold music on a smartphone:
• 1950s valve radio:
• small speakers in a large hall, so compromised sound:
• last night at dance class, on a portable Bluetooth speaker:
• my sister's Echo Dot (she has a decent setup as well, but her friends' seasonal radio broadcasts are on low bitrate internet and AM radio, so she listens to them on it because she wants that sound for that experience):
• three or four decent hifi systems, including my own

So, in none of those cases, even though audio was sometimes quite badly compromised, was spoken voice unintelligible, timing of the start of instrument sounds affected (the ends were, due mainly to spaces the devices were in, as I heard it), I could tap my foot or dance to the music concerned, I could identify different individual instruments quite happily.

I'm not arguing that these devices were equivalent in overall audio quality, far from it. I'm just saying that the transient response of drive units, even older ones, were all quite capable in terms of starting and stopping, and producing what they were designed to produce according to the space they were in. Of course the hifi systems did better: they had better frequency range and response, almost certainly had better sources, they were louder (never underestimate that), Transient response, in terms of starting and stopping diaphragm movement? Hardly an issue in comparison with all of the other shortcomings.

I've reported another experience here before, where a smartphone outperformed a half-million dollar but badly set up system in a store demo room, for the really important details of an orchestral recording. While I'm not going to go as far as to say that speaker transient response is completely "solved" or anything that dumb, it's hardly the problem that bad design and setup, poor frequency response, high distortion and the other usual suspects are.

 

[theREALdotnet]

We're talking speakers here though. The speakers' particular ability doesn't change because the setup of the rest of the system isn't perfect. And I can assure you that low cost speakers from the 1980s can achieve that sort of effect on occasion.

No, you’re right, on both accounts. No single component’s ability can make up for another‘s failure. The faults add up, the virtues don’t. And yes, low-cost speakers from the 80s is what I’m using, too. I bought mine in the early 90s and have not found anything that would make me want to spend money to replace them.

I'm going to stay on the very controversial edge of this though. When I have heard some of these startling effects and oddities, they aren't "real", and I can be pretty sure that they are mostly not part of what was intended in the recording. Singers are supposed to be on stage, drummers normally at the back, guitars to the sides, and so on. First violins to the left, second to the right, cellos to left of centre, (maybe in front and slightly to the left or well to the left according to the orchestra's practice), and so on. When an instrument suddenly jumps to sound "realistic" and way off to the side, or the bass player "sits in my lap" (yes, I've had that appear to happen) I regard it as wrong.

Perhaps I wasn’t clear in my description – has anyone ever been able to describe an auditory impression on ASR without being laughed out of the room?

The startling bit I was trying to convey is not an effect but a property of the (stereo, not multi-channel!) playback system. It requires careful sound recording of an actual acoustic performance, or it can be concocted, with an equal amount of care, at the controls of some marvelous piece of technology. My point was, even if the recording contains it, it is easily spoiled. In fact, 9 out of 10 systems I’ve listened to in my lifetime have not been able to reproduce the sense of realism in recordings that I knew had it in them.

I tend to think of two channel stereo (not so surround, if that is what you have) as being a bit like a slightly 3D TV image. Things should be in place and I should be able to understand everything in context. Even if you have a strong central image for your vocalist, if you have a choir or operatic chorus involved, or different singers placed across the soundstage, that one sticking out becomes, well, a problem.

Yeah, TV/movie sound is of no particular interest to me, I make do with a simple 4.0 system there that keeps the family happy. My obsession is with my stereo setup.

So, in none of those cases, even though audio was sometimes quite badly compromised, was spoken voice unintelligible, timing of the start of instrument sounds affected (the ends were, due mainly to spaces the devices were in, as I heard it), I could tap my foot or dance to the music concerned, I could identify different individual instruments quite happily.

With the right music playing I can even listen contently to AM radio. The kind of fidelity we’re discussing is not about bobbing your head to a song, or understanding what a person says, it‘s about the difference between a really nice sounding stereo system and a convincing 3D auditory illusion – from a pair of stereo speakers. I’ve had friends over to whom I played some pop and rock (Madonna and AC/DC, nothing fancy), who thought I had speakers hidden in all those acoustic panels in my room.

I wish I knew where this quality comes from. It’s likely a mix of qualities, but I believe that phase and transient response have something to do with it. If only for the fact that those are never measured, and none of measurements taken can seemingly explain the differences between loudspeakers (at the level I described above).

 

[Tangband]

Just curious, but hasn't the square wave/step response matter in traditional x-overs been dealt with in numerous threads and referred to as inaudible?

I’m sure it has. But so has the matter of stereo being silly and mono all we need (or true multi-channel), or the matter of an amplifier being state-of-the-art if it survives a week.

Kidding aside, there are listening demos which clearly show that phase distortion is ignored at your own peril, and square/step/impulse responses are simply ways of showing said distortions more easily. The thing with phase response is, sometimes accuracy matters, sometimes it doesn’t. Same with magnitude response, funnily.

Everything matters . Ofcourse stepresponse and phase differences matters, otherwise we couldnt experience stereo .
The distance between our two ears are less than 1 ms soundtravel. So we hear phase differences in a very short distance.
Within the first 1-2 ms ( 34- 68 cm ) our brain can locate where the sound is coming from. The brain locks on the first incoming wavefront, and attenuate sound reflexes from the sidewalls with up to 10 dB if the room is big. This also means that the direct sound on axis from a loudspeaker is more important than the off axis sound, in a big room.

 

[theREALdotnet]

Everything matters . Ofcourse stepresponse and phase differences matters, otherwise we couldnt experience stereo .
Within the first 1-2 ms ( 34- 68 cm ) or brain can locate where the sound is coming from.

Yes, everything matters. Phase response (like magnitude response) is a real thing, i.e the behaviour of an actual device in the physical world. Step response, impulse response, FR plots etc. are just ways for us humans to measure it and visualise it on a graph.

We seem to pay a lot of attention to how a speaker radiates its energy into the room, so we take spinoramas, predict room interaction and all that. This is of course very important for characterising a speaker’s in-room performance, but equally important is the stimulus that comes out of it and excites the air in the first place. And this is not sufficiently characterised by a Klippel plot, containing only steady-state magnitude response information (as far as I can tell).

 

[Thomas_A]

Everything matters . Ofcourse stepresponse and phase differences matters, otherwise we couldnt experience stereo .
The distance between our two ears are less than 1 ms soundtravel. So we hear phase differences in a very short distance.
Within the first 1-2 ms ( 34- 68 cm ) our brain can locate where the sound is coming from. The brain locks on the first incoming wavefront, and attenuate sound reflexes from the sidewalls with up to 10 dB if the room is big. This also means that the direct sound on axis from a loudspeaker is more important than the off axis sound, in a big room.

Stereo is another thing. A classic x-over and its distorted step response is not audible.

 

[Tangband]

Stereo is another thing. A classic x-over and its distorted step response is not audible.

You need to do a couple of years with experimenting with digital crossovers ( as I have done ) to see If your statement is true . It isnt, not in all cases.

 

[ernestcarl]

Stereo is another thing. A classic x-over and its distorted step response is not audible.

Not casually, no.

I have uploaded an ABX I did a while back where I did hear something through my KH120s.

It’s not particularly significant enough for me personally to seek a “perfect step” in all/most cases — although, I’d rather have the better sounding system i.e. better transient and phase as well. But the “center fill” channel I use before (Fostex 6301) and now Fulcrum RX699 have a different response. Some amount of FR matching is required, phase response included, obviously. No point introducing the same phase distortion to the single driver or coax center.


6301 mounted on screen

RX699 mounted on wall

 

[levimax]

View attachment 288270

then you can kind of guess that the speaker designer didn’t pay enough attention to phase response. Any composite signal (like music) that undergoes phase distortions like this will probably sound less real or convincing to our ears. This may also affect sound stage and imaging, who knows. There doesn’t seem to be enough research on this.

As much as this "looks bad" there really is no evidence (and there has been a fair amount of research) that this "sounds bad". Intuition can be useful but not always and I think phase and square waves are examples of things where it makes intuitive sense that good square wave reproduction and flat phase should sound better but there is little evidence this is the case and a lot of evidence that it doesn't make any audible difference. This is backed up by my experience with REW and Rephase to flatten phase.... I can measure and see the "success" of the phase correction but if I ABX with and without the phase correction I can not tell a difference. I think this is valuable information because it prevents me from wasting time and money on things that don't matter.

 

[Tangband]

Yes, everything matters. Phase response (like magnitude response) is a real thing, i.e the behaviour of an actual device in the physical world. Step response, impulse response, FR plots etc. are just ways for us humans to measure it and visualise it on a graph.

We seem to pay a lot of attention to how a speaker radiates its energy into the room, so we take spinoramas, predict room interaction and all that. This is of course very important for characterising a speaker’s in-room performance, but equally important is the stimulus that comes out of it and excites the air in the first place. And this is not sufficiently characterised by a Klippel plot, containing only steady-state magnitude response information (as far as I can tell).

True. The interesting fact is also that noone has explained in detail how the perceived phantom stereo experience is created inside the brain - this is far more complicated than any measurements can explain. We dont know everything - yet.

 

[Holdt]

True. The interesting fact is also that noone has explained in detail how the perceived phantom stereo experience is created inside the brain - this is far more complicated than any measurements can explain. We dont know everything - yet.

Maybe it's also different from ear to ear. I mean not all ears look the same so it's plausible to assume that the sound we hear and perceive can be different? At least IR to this 3D phenomenon that is phantom center and perceived surround from two speakers.

 

[Tangband]

Maybe it's also different from ear to ear. I mean not all ears look the same so it's plausible to assume that the sound we hear and perceive can be different? At least IR to this 3D phenomenon that is phantom center and perceived surround from two speakers.

Yes, we humans are physically slightly different. Also, music is a language that each of us can learn, and we are at different levels and that fact alone makes us experience and enjoy music in very different ways. Some people like advanced Jazz music, and for some people thats just noise.
Whats common for most people is that most of us can hear and enjoy a tuned guitar and that most of us dont like a guitar or piano thats out of tune. We can also feel rythm, and can hear if a bunch of musicians are playing together , or not .

 

[Thomas_A]

Not casually, no.

I have uploaded an ABX I did a while back where I did hear something through my KH120s.

It’s not particularly significant enough for me personally to seek a “perfect step” in all/most cases — although, I’d rather have the better sounding system i.e. better transient and phase as well. But the “center fill” channel I use before (Fostex 6301) and now Fulcrum RX699 have a different response. Some amount of FR matching is required, phase response included, obviously. No point introducing the same phase distortion to the single driver or coax center.

I have not seen any experimments/ABX that show audibility of phase shift of eg LR4 @ 3 kHz. Can you link if so?

 

[Tim Link]

True. The interesting fact is also that noone has explained in detail how the perceived phantom stereo experience is created inside the brain - this is far more complicated than any measurements can explain. We dont know everything - yet.

Is anything explained in detail yet about how perceptions are formed in the brain?

 

[ernestcarl]

I have not seen any experimments/ABX that show audibility of phase shift of eg LR4 @ 3 kHz. Can you link if so?

If you mean to say formal experimental papers, I don't know as I do not follow this closely enough. My personally conducted ABX listening test was referring to phase shift at around 2 kHz from said KH120s. Well, I can also hear the phase shift of the uncorrected passive crossover from my coax RX699 at 2 meters. But... I guess it doesn't exist since there is no formal proof?

 

[Thomas_A]

If you mean to say formal experimental papers, I don't know as I do not follow this closely enough. My personally conducted ABX listening test was referring to phase shift at around 2 kHz from said KH120s. Well, I can also hear the phase shift of the uncorrected passive crossover from my coax RX699 at 2 meters. But... I guess it doesn't exist since there is no formal proof?

Phase shifts can be heard if they are big enough. I was mostly interested in the phase shifts caused by the traditional passive crossovers in the 1-3 kHz range. ABX is enough if it is demonstrated with the usual result dumps.

 

[Cbdb2]

It is a fact that music does not contain anything like a square wave so what would be the point of worrying about reproducing one? If a speaker can reproduce 20 Hz to 20 Khz at a linear SPL that is all you need for listening to music. Chasing things that don't make a difference is a waste of time and money and learning what is important and what is not important is the value this site offers vs the "everything matters" audiophile myths.

Synthezisers.

 

[levimax]

Synthezisers.

Yea I was thinking about that as well. It is kind of an interesting thing to think about as even if a synthesizer is creating a close to proper square wave if they are high enough frequency they are not going to be accurately recorded with a bandwidth limited digital recording. That is OK though because a square wave will sound like a square wave as long as all the audible harmonics are accurately recorded. If harmonics beyond human hearing are "lost" during the recording process it will mess up the "graph" of the square wave but it won't affect how we perceive it.

 

[Cbdb2]

They are always an approximation, and get LP filtered even with analog gear. But there's nothing like a 40hz sq wave to get all the drivers going.

 

[Galliardist]

Synthezisers.

I’ve yet to hear an electronic instrument that doesn’t itself require a loudspeaker, or something analogous to a traditional instrument, to make a sound. That square wave is always different in practice.

Of course, it is differently reproduced in our systems, but it is never heard as a pure square wave, anywhere!

 

[levimax]

Of course, it is differently reproduced in our systems, but it is never heard as a pure square wave, anywhere!

Yes, and they don't sound very nice either.