Measuring transient response?

[ppataki]

I was wondering how it would be possible to measure the transient response of a loudspeaker using a measurement microphone (e.g. UMIK-1) and a software (e.g. REW)
Here is the idea I came up with: I have generated 1ms sine tones at various frequencies at a certain volume and the same at a volume 30dB lower
Then I play these right after each other, first the loud one then the quiet one and I measure the response with the Logger feature of the SPL meter in REW (using C weighting and Fast mode)
I have generated these tones at 30, 40, 100, 440, 1000 and 7000Hz and I have also done the same with full range and limited range (500-2000Hz) pink noise
In all measurements I could see a dynamic range between 25 and 29dB so pretty close to the 30dB difference
Some examples:

40Hz

1000Hz

PN full range

I would like to ask if this makes sense and if there are better ways to measure the same?

Many thanks in advance

 

[Blumlein 88]

I think @RayDunzl used like a 3hz square wave. Effectively each leading edge is a steep transient.

Now you are wasting your time. The transient or impulse response can be calculated from the FR sweep in REW. Ray was using the square wave to show you get the same thing either way.

 

[ppataki]

Thank you @Blumlein 88

Here are two impulse response measurements of my two systems, can you please help me understand how to interpret the transient response out of these?

 

[tuga]

I wonder if this is of any help:

https://www.stereophile.com/features/100/index.html

An impulse response is extremely hard to interpret, not least because, with a loudspeaker, it is visually dominated by the tweeter's output.
Its shape doesn't really tell you much in itself about woofer and midrange-unit behavior.

 

[RayDunzl]

I think @RayDunzl used like a 3hz square wave. Effectively each leading edge is a steep transient.

What I wanted to do was to compare the Impulse and Step response that REW calculates from the sine sweep to an Impulse (single full scale sample) and Step (using a 10hz square wave - 50ms high, 50ms low).

That experiment was surprisingly successful.

https://www.audiosciencereview.com/forum/index.php?threads/impulse-response.1765/#post-44352

https://www.audiosciencereview.com/forum/index.php?threads/impulse-response.1765/#post-44440

---

I was wondering how it would be possible to measure the transient response of a loudspeaker using a measurement microphone (e.g. UMIK-1) and a software (e.g. REW)

Transients:

Musical transients aren't so clear cut as an Impulse or Step.

Below is a single Guitar pluck courtesy Al di Meola.

The three tracks are the (converted to mono) source from CD, recorded playback (at the listening position) using my big speakers, and recorded playback of my little speakers.

Note that room reflections soon change the waveform from what might be expected.

See milliseconds at the top.

Also, I'm surprised the JBL looks so much "worse", but they do excite many more refelctions in the room..

No EQ was in use:

"Perfection", in my opinion, would be "all three look exactly the same. Doesn't happen.

What the above is looking at was this "transient"

 

[Blumlein 88]

Thank you @Blumlein 88

Here are two impulse response measurements of my two systems, can you please help me understand how to interpret the transient response out of these?

View attachment 90977

View attachment 90978

I'm not sure what to say other than it doesn't look bad. If you read the stereophile link provided by tuga it explains what you see somewhat. Were these measured at your listening position? To see what the speaker alone is contributing it is a good idea to measure close to the speaker.

 

[tvrgeek]

http://audiojudgement.com/cumulative-spectral-decay-csd-plot/

 

[RayDunzl]

I'm not sure what to say other than it doesn't look bad.

Well, 1+ 1 does not necessarily equal 2 when you have a source signal and a playback system in a room.

If you read the stereophile link provided by tuga it explains what you see somewhat.

There's a lot going on.

Were these measured at your listening position?

Yes.

That's where I listen, if listening attentively.

The rest of the listening time I'm at the desk in the right rear corner (now), or at the napping position, head at the left end of the couch.

The impulse and step response for the napping position is something like this:

I can delay the left speaker by 1.7ms at the miniDSP and get the stereo back, but typically don't, since stereo isn't that obvious when one ear is on the pillow, making the East-West Intra-Aural Time Difference close to zero.

We're all used to listening to stereo off-axis anyway.

To see what the speaker alone is contributing it is a good idea to measure close to the speaker.

I could with the JBLs. They "clean up" nicely.

"Measuring panel speakers is fraught with difficulties", to paraphrase @John Atkinson.

 

[andreasmaaan]

@ppataki what are you hoping to learn from these measurements? Knowing this would make it easier to give you some advice about which measurements to take, how to process/transform them, etc.

 

[Few]

While I have a longstanding fascination with time domain speaker measurements, they can be tricky to interpret. Things that look awful might do so primarily because of a phase response anomaly that is not as audible as the impulse or step response might lead you to believe. In other words, there can be a large disconnect between how bad things look and how bad they sound. On the other hand, a nearly perfect impulse response implies a nearly perfect magnitude and phase response, so I suppose you could say if it looks good in time it is good.

I wrote a program long ago that would allow me to examine impulse and step responses that would result from crossovers that yielded flat magnitude response (SPL) but not linear phase. I found it quite educational. Sorry I don’t have it any more to share! Maybe someone can suggest a readily available alternative.

Few

 

[ppataki]

@andreasmaaan Simply I just would like to see how good the system's transient response is
I have been doing frequency, phase, impulse, etc. measurements but I am struggling how to have an idea about transient response just by looking at a measurement
(if that is possible at all...)