Step Response: Does It Really Matter?

[j_j]

I wonder how you do that.

Well you can start with the impulse response. Go to www.aes.org/sections/pnw and then in "meeting recaps" or something like that to tell you how.

if you have the impulse response, you can just sum along the axis adding one sample at a time, there's your step response.

Note; using an actual impulse is a bad idea. Use an allpass sequence. Look at the site, find the talk, you'll see. There is even a test signal and a matlab script there for you to use if you have that, or instructions on how to get octave if you don't spend for matlab.

 

[j_j]

"Microphones work on the principle of pressure, pressure gradient, or velocity. If you don't know the difference, then you can't yet call yourself a sound engineer."

Whoa, dodged that bullet...

More at https://www.audiomasterclass.com/ne...re-pressure-gradient-and-velocity-microphones

There are inexpensive soundfield mikes out there. Just use one. While ambisonics is NOT the holy grail of stereo capture despite what its proponents think it DOES capture the entire soundfield at one point.

 

[DonH56]

I haven't tried to do pressure/velocity measurements for ages. When I did, I used my old B&K or Sennheiser (forgot model, their equivalent of the Earthworks M30 omni measurement mic I have now) for the pressure mic and a large-frame condenser (but cheap, e.g. AKG C3000) for the "velocity" sensor since I didn't have a good ribbon mic at the time. I did borrow a friend's ribbon and IIRC the results weren't much different (a ribbon is not a true velocity mic but about as close as it gets). I have a vague memory of papers trying to use the plasma driver approach in reverse as a mic to achieve a true velocity mic, and even got pinged on the idea back in the early 1990's to help a startup working on the approach, but it never really became a commercial success AFAIK.

I do not recall the results ever being that useful except as a laboratory exercise; maybe @j_j could comment on the practical use? I graduated and moved on shortly after doing the measurements (for a grad class project) away from audio research.

 

[j_j]

I do not recall the results ever being that useful except as a laboratory exercise; maybe @j_j could comment on the practical use? I graduated and moved on shortly after doing the measurements (for a grad class project) away from audio research.

If you're trying to measure acoustic impedance, it's very useful. This is somewhat esoteric, unless you build mikes, or speakers, or measure soundfields.

 

[DonH56]

If you're trying to measure acoustic impedance, it's very useful. This is somewhat esoteric, unless you build mikes, or speakers, or measure soundfields.

Yes, and that's what I was doing at the time, just was not sure of it's application to consumer measurements. Does not seem like something the average audiophile needs to muck with (or worry about)...

Thanks JJ! - Don

 

[MattHooper]

I agree. Someone else who says the same thing:

I can unfortunately add nothing of technical interest to this thread, and even though I own Thiel (time/phase coherent) speakers I remain agnostic about the case for time/phase coherency, given I see arguments from knowledgeable people on both sides of the debate.

But I can't help remark on the paper above. Many of the listening impressions purported for time/phase coherence seems to align so well with what I hear from the Thiel speakers (2.7 with coax mid/tweeter).

One really stuck out in the paper:

"3.Separation of ambience. With loudspeakers whose stereo image is slightly blended because of time-smear, any hall ambience or reverberation in the recording tends to become slightly mixed with the instrumental sounds, causing coloration of those sounds. Consequently, with such speakers closely-microphoned recordings tend to sound better because of their distinctly defined sound. But with time-corrected loudspeakers, the ambience is resolved as a separate sound, and larger amounts of hall ambience in recordings can be enjoyed.......”

That is EXACTLY one of the attributes that stick out every time I compare my Thiels directly to my other speakers - MBL omnis, Waveform Mach MC, Spendor, Hales and others - or to other non time/phase coherent designs.

Recently I've been switching in a variety of my other speakers and have truly enjoyed them - each brings something I like, sometimes that aspect a bit more than the Thiels. But what always sticks out is the relative lack of imaging precision. It's not like my other speakers don't soundstage and image impressively - they do so like gangbusters. But I can't help but note how the sonic image of a voice, sax or any other instrument on the other speakers seems to be blurred or mixed in with the surrounding reverb or acoustic. It gives a sort of flattened or see-through quality to the sonic images. Every time I go back to the Thiels, images seem to "snap" in to focus, as if all the sonic energy coming from a sound source which goes astray in other designs has been organized and lined up properly. The effect is that sonic images seem more solid, dense, dimensionally 'round' and separated from the reverb and acoustics. So they sound like solid acoustic sources IN a reverberant space.

Again, I can't speak to whether this is due to their time/phase coherence, or perhaps to some other attribute of their engineering. But it is a constant impression I have every time I go from listening to a higher order non time/phase coherent design to the Thiels.

 

[MattHooper]

Also, I have to hop on the earlier conversations about electrostatic speakers. One poster said he'd moved on from them, finding them dynamically unconvincing - not 'pressurizing the room' the way box/dynamic speakers do. I moved on from electrostatics (Quade ESL 63s) for the same reason! I was for a while seduced by that particular boxless 'transparent-sounding' aspect of the electrostatic sound. But it seemed after a while to have a sort of ghostly, removed character, as if I were listening to a performance happening through a window in a different room.
When I'd throw in my older smaller box speakers (a pair of old Thiel 02s) it's like the room came alive with the energy of musicians. Drums, bongos just seemed to "whap!" the air right there, like someone was in the room playing the instrument. It just seemed more palpable and dynamic and affecting.

I added the gradient sub to the Quads - some may remember it was a dipole sub designed specifically to match the radiation pattern of the Quads, and the Quads sat on top making for a seamless monolith. They remain to my ears the most seamless combination of dynamic sub with a panel that I've heard. Yet, I still found the 'problem' above persisted and I moved on to box speakers, never looking back.

I've heard a great many Martin Logan hybrids over the years, and my friend owns a pair of ML hybrids as well. Every time I'm struck by the same impression: wow at that cool electrostatic presentation in the mids up, but that same old weightlessness. I get why some people think that adding a dynamic driver in to the design for the lower frequencies gives back some dynamic palpability. But what I always hear is a discontinuity: instruments ranges covered by the dynamic driver have some weight and air-moving drive - e.g. bass instruments - but as the range moves to higher frequencies there is that weightless quality. My friend thinks his MLs rock just fine because once he feels those bass notes hitting from the woofer section, well....there you go! They rock! But for me, it just doesn't work as I find the rest of the spectrum dynamically bereft. Put on Rush and geddy lee's bass can be felt, but Lifeson's guitars don't seem to have the same drive - they are more like a sonic mist you could just walk through rather than the killer impact you hear from a guitar through their cabinet amplification.

So, I get both the anti-panel and pro panel opinions. I enjoy visiting panels - and in fact would own Quad ESL 57s if I had the room to store them when not listening. But it's more a nice place to visit than a place to stay for me.



 

[bennybbbx]

I think step reponse is usefull which time it need to reach 0 and how much % it later reach max to see in this value if a speaker is precise and can produce good stereo width or not.

in this tests
https://www.audiosciencereview.com/forum/index.php?threads/test-ikmultimedia-iloud-mtm.19814/

https://www.audiosciencereview.com/...step-response-and-audio-record-example.19813/

https://www.audiosciencereview.com/...tep-response-and-audio-record-examples.19812/

i can clear hear that on kali LP6 stereo width is much smaller and sept response is very worse. correct the phase help a little, but it is not much that is need. the record of the speaker is phase exakt, because i record first left then right channel on same speaker and mikrophone in DAW. there should hear that 2 guitars play. 1 guitar come from left speaker position 1 guitar come from right speaker position. but with the kali with slow transient i hear that the guitar come more in the middle and not from left speaker or right speaker positiion. maybe other can too do speaker records to have more confirmation that slow step response give reduce stereo width because of unprecise speaker.

test me and when i choose which speaker have better stereo width, then you can show step response how it look

ears are able to hear 10-20 microseconds diffrence. and stereo hearing is more in mid range. so it seem important too that mid range have fast precise speakers.

Abstract

The auditory system encodes the timing of peaks in basilar-membrane motion with exquisite precision, and perceptual models of binaural processing indicate that the limit of temporal resolution in humans is as little as 10-20 microseconds. In these binaural studies, pairs of continuous sounds with microsecond differences are presented simultaneously, one sound to each ear. In this paper, a monaural masking experiment is described in which pairs of continuous sounds with microsecond time differences were combined and presented to both ears. The stimuli were matched in terms of the excitation patterns they produced, and a perceptual model of monaural processing indicates that the limit of temporal resolution in this case is similar to that in the binaural system.

https://pubmed.ncbi.nlm.nih.gov/12765396/


i hear some weeks ago loud music longer time. And then happen that i hear all in less stereo width and i feel pressure on ears. i think my ears are overload and after a rest all was ok. so it can maybe happen when people hear long and loud music, that ears maybe reduce in binbaural hear precision. that frequency loss happen on ears when people hear often loud all know, but it is possible that binaural hear loss happen too because 22 microseconds are 44 khz.

so even if hifi experts do not hear diffrence on slow or fast step response, people that hear not often loud music or are young can hear this. I was 1 time in my life on a concert. it was too loud for me and my ears feel too bad after this. all sound strange but not high frequency loss. now that i do more about stereo width tests i think i know whats happen. l loose ability of binaural hearing temporarly. and because i never go to concert or hear loud to avoid this feeling i maybe hear over 50 ok stereo.

 

[j_j]

https://pubmed.ncbi.nlm.nih.gov/12765396/

I think that to some extent you're missing an extremely important part, the matching between speakers. This devolves instantly to crossover tolerances.

For arrival times at different frequencies, it seems quite obvious and clear to me that the real question is impulse response.

As to stereo width, there are a variety of issues far, far beyond loudspeakers and the playback chain.

But your "22 microsecond" comment is pure nonsense. The time resolution of 44/16 is not 22 microseconds, or anything near that. It is, more precisely, 1/(20000*2*pi*65536). This is very much smaller than 22 microseconds. You've just restated a common, and extremely inaccurate, myth.

 

[j_j]

but it is possible that binaural hear loss happen too because 22 microseconds are 44 khz.

This association is completely mistaken.

 

[bennybbbx]

I think that to some extent you're missing an extremely important part, the matching between speakers. This devolves instantly to crossover tolerances.

For arrival times at different frequencies, it seems quite obvious and clear to me that the real question is impulse response.

As to stereo width, there are a variety of issues far, far beyond loudspeakers and the playback chain.

But your "22 microsecond" comment is pure nonsense. The time resolution of 44/16 is not 22 microseconds, or anything near that. It is, more precisely, 1/(20000*2*pi*65536). This is very much smaller than 22 microseconds. You've just restated a common, and extremely inaccurate, myth.

In my audio record examples i use same speaker and same microphone on same place. I record first left channel. then right channel. because i use a daw the records are sample sync.

i have calc the period time with this when i insert 44100 hz the result is exactly 22.675737 microseconds. so my 22 microseconds post is not so much diffrence and the 0.6 microsecond diffrence is not important

https://www.sensorsone.com/frequency-to-period-calculator/

 

[tuga]

Joe D’Appolito talks about step response here:

Step Response

Up to this point we have looked at loudspeaker performance solely in the frequency domain. Let’s turn now to the time domain for additional performance insight. We could examine the impulse response in more detail, but it is not easily interpreted. It is dominated by the tweeter response in the first few milliseconds. It doesn’t tell us much about the woofer, or the midrange if there is one, because all the low-frequency information is in the impulse response tail, which is at a very low signal level. The step response is a much more useful tool.

The step input is a signal that rises instantaneously from zero to a fixed level. This is basically a DC input starting at time zero. Mathematically, the step response is the time integral of the impulse response.

Continues below

https://audioxpress.com/article/testing-loudspeakers-which-measurements-matter-part-2

 

[bennybbbx]

here are headphone step response of some cheap headphones and a expensive but with large membrane. you can see in compare it reach 0% much much faster and much less overshoot

 

[j_j]

i have calc the period time with this when i insert 44100 hz the result is exactly 22.675737 microseconds. so my 22 microseconds post is not so much diffrence and the 0.6 microsecond diffrence is not important

The per-channel and interchannel time resolution of PCM audio is MUCH SMALLER than 1/sampling_rate. The correct example is given above in my reply. Please do not repeat the false myth that the time resolution of PCM is 1/sampling_rate, because that is extraordinarily wrong.

 

[j_j]

here are headphone step response of some cheap headphones and a expensive but with large membrane. you can see in compare it reach 0% much much faster and much less overshoot

Now please calculate the impulse response, and look at the envelope of the impulse response. Step response is not the best measure here.

Also, you must use simultaneous sampling in left and right channels, and you need to look at the per-1/3 ERB envelope in the two channels, compared across the two channels.

 

[bennybbbx]

Now please calculate the impulse response, and look at the envelope of the impulse response. Step response is not the best measure here.

Also, you must use simultaneous sampling in left and right channels in previous post from me, and you need to look at the per-1/3 ERB envelope in the two channels, compared across the two channels.

the step response is calculate in the left and right screenshot. in the middle it is impulse response of several measurements so no can say wrong measure. in the middle screenshot i use diffrent microphone positions but impulse look same. i have also do step response from 300 hz up of speaker. they are also slow

 

[j_j]

you are wrong. when want delay a signal with a DSP effect, then you can only delay in sample steps. and 1 sample step is 22,6 microseconds. and when hear a signal with 22 khz then period time is 45.2 microseconds. people can not hear this, but can hear delay of 45 microseconds between left and right ear easy

Ok. One of us does this for a living. That's me. Subsample delays are well understood, easily implemented, and documented time and time again in the literature. Your insistence on this myth is pure, 100% disinformation, and has been presented over and over, with the result that the myth of intersample resolution persists, as your false information proves conclusively.

Just give it up. You do not understand sampling, sampling theory, or signal reconstruction theory. You just do not understand what you're talking about.

http://users.spa.aalto.fi/vpv/publications/vesan_vaitos/ch3_pt1_fir.pdf

Don't let the "approximation" comment fool you, the approximation can be to any arbitrary level of accuracy that one might want.

https://www.intechopen.com/books/ap...-engineering/fractional-delay-digital-filters Read that for (*&*(&( sake instead of making elementary mistakes.

https://ieeexplore.ieee.org/document/860248 for more modern treatment.

But let me help you. Take your 44/16 signal, interpolate it to 88/16. This kind of sampling rate conversion is well known. Then, delay the 88kHz signal by 1 sample. Now, convert back to 44/16. Now you have 1/2 sample delay, to any arbitrary precision that you might wish.

Or, maybe interpolate by 128, then interpolate adjacent samples linearly, then decimate by 128. Now you can do fractional delays to the resolution of your floating point processing.

Q.E.D.

 

[pma]

If you're trying to measure acoustic impedance, it's very useful.

Attached is a part of the paper (Impulse measurements of electrical and acoustical impedance) by my university lecturer, professor Josef Merhaut. Maybe you have known him?

https://asa.scitation.org/doi/pdf/10.1121/1.387330

 

[j_j]

the step response is calculate in the left and right screenshot. in the middle it is impulse response of several measurements so no can say wrong measure. in the middle screenshot i use diffrent microphone positions but impulse look same. i have also do step response from 300 hz up of speaker. they are also slow

Now calculate the analytic envelope of your impulse response. Do so in each ERB. Look at the time delays. Learn.

https://dsp.stackexchange.com/questions/424/hilbert-transform-to-compute-signal-envelope

For two channels, keep sampling running so it is sample-synched. Capture your probe signal in another channel so you can verify your timing is precise.

Then use the analytic envelope. Plot the two channels on one plot. Look at the accuracy. Use octave if you need a signalprocessing engine.

 

[j_j]

Attached is a part of the paper (Impulse measurements of electrical and acoustical impedance) by my university lecturer, professor Josef Merhaut. Maybe you have known him?

https://asa.scitation.org/doi/pdf/10.1121/1.387330

This has nothing to do with time resolution of PCM or subsample delays. It may read (if the detail is there) on how to calculate acoustic impedance, at least for the English abstracts. It does appear that the paper has some discussion of acoustic impedance. I suspect that there may be a SNR argument that can be avoided in more modern methods, but I can't read the actual paper, only the abstract.

It is important to understand that unlike what some posters are doing here (with tone bursts) impulse response must be calculated with broadband stimulii. Post-analysis can easily then apply any frequency sorting necessary.