Speaker "Speed"

[Piranesi]

Apologies in advance if this turns out to be a silly question.

Are there any metrics for the transient response or "speed" of a speaker? Subjectively, certain ribbon speakers I've heard have reproduced cymbals and strings with a faster attack than otherwise comparable tweeters.

I've seen it being described on other forums as the ability to accurately follow a square wave, but I've no idea if this is an accurate definition. I'm also not sure if the relevant metrics are already present in Amir's measurements.

 

[ppataki]

I also asked this question here:
https://www.audiosciencereview.com/forum/index.php?threads/measuring-transient-response.17219

But no conclusion so far, still not sure how to measure transient response

 

[detlev24]

Are we talking about the same "speed" that usually is referred to in 'high-quality' subwoofers, like "fast" drivers?

If so, this is the ability of a driver to stop its impulse. Usually not a problem with small drivers due to a relatively low mass, but certainly something to take into consideration with big drivers. For simplification only, I would refer to "small" woofers as ≤ 10" in size and to "big" woofers as > 10" in size.

Rythmik Audio, for example, call their solution 'Direct Servo', which is nothing but a "stop motor": an additional sense coil that gets its feedback through the amplifier without delay, to stop any given movement. The effect could be measured via group delay; and will manifest in magnitude of distortion.

====
Edit: A similar phenomenon could also be the reason in lower quality tweeters.

 

[andreasmaaan]

I also asked this question here:
https://www.audiosciencereview.com/forum/index.php?threads/measuring-transient-response.17219

But no conclusion so far, still not sure how to measure transient response

I didn’t read the thread, but there should be a clear conclusion!

Measure the impulse response (or any signal that can be transformed into an impulse response, eg a sweep). That measurement will contain a full characterisation of your loudspeaker’s transient response (within the limits of your measurement setup, ofc).

It’s really that simple

 

[Piranesi]

I didn’t read the thread, but there should be a clear conclusion!

Measure the impulse response (or any signal that can be transformed into an impulse response, eg a sweep). That measurement will contain a full characterisation of your loudspeaker’s transient response (within the limits of your measurement setup, ofc).

It’s really that simple

Wouldn't this be worth measuring then? It seems like it'd be a fairly significant contributor to perceived sound quality.

 

[hmt]

Apologies in advance if this turns out to be a silly question.

Are there any metrics for the transient response or "speed" of a speaker? Subjectively, certain ribbon speakers I've heard have reproduced cymbals and strings with a faster attack than otherwise comparable tweeters.

I've seen it being described on other forums as the ability to accurately follow a square wave, but I've no idea if this is an accurate definition. I'm also not sure if the relevant metrics are already present in Amir's measurements.

No speaker with minimum phase crossovers can follow a square wave accurately. Faster attack is usually pointed to speakers with a bright signature which many speakerw with AMT and the likes show.

 

[MZKM]

The effect could be measured via group delay;

Isn’t group delay the time it takes to start playing (“lag”), not a measure of how it plays it for (“hang time”)?

 

[MZKM]

I've seen it being described on other forums as the ability to accurately follow a square wave, but I've no idea if this is an accurate definition.

Just like with amps, this isn’t a good metric.
The “rise time” is influenced a lot by the high frequency extension. The square wave on speaker that can only play up to 20kHz will never look as good as one that can play to 40kHz.

 

[detlev24]

Isn’t group delay the time it takes to start playing (“lag”), not a measure of how it plays it for (“hang time”)?

Group delay is the rate of change between phase shift vs. frequency.


The following might be of academic interest, concerning the audibility of time offset(s) without frequency response aberrations:

Some Experiments With Time: On the audibility of phase shift (by David Clark)

 

[MZKM]

Group delay is the rate of change between phase shift vs. frequency.


The following might be of academic interest, concerning the audibility of time offset(s) without frequency response aberrations:

Some Experiments With Time: On the audibility of phase shift (by David Clark)

Which is a time delay, no?

 

[aarons915]

Apologies in advance if this turns out to be a silly question.

Are there any metrics for the transient response or "speed" of a speaker? Subjectively, certain ribbon speakers I've heard have reproduced cymbals and strings with a faster attack than otherwise comparable tweeters.

I've seen it being described on other forums as the ability to accurately follow a square wave, but I've no idea if this is an accurate definition. I'm also not sure if the relevant metrics are already present in Amir's measurements.

It's not a silly question but just like the myth of "fast woofers", frequency response seems to characterize what we hear in woofers and tweeters. The frequency response is derived from the impulse response and excess decay in the impulse usually shows as a peak in the frequency response as well. Another argument for ribbon tweeters is the idea of them being very low mass contributing to this "speed" but a tweeter with more mass will have a larger motor to keep the acceleration the same, basically F=ma.

 

[okok]

metal drivers?
smps amps?
output z really low?

 

[andreasmaaan]

Wouldn't this be worth measuring then? It seems like it'd be a fairly significant contributor to perceived sound quality.

It's always measured, the question is whether it's worth displaying the results!

In general, I don't think it would hurt to display them (e.g. as a phase response graph, a step response, or a group delay graph - all of which are alternative methods of representing the same information).

However, the research suggests that it is rare that the relatively low degree of this type of distortion (which is usually referred to in the psychoacoustics literature as group delay) exceeds thresholds of audibility in typical loudspeakers. I believe this is why the information is usually omitted.

If you're interested in hearing what the effects of this type of distortion sound like (if indeed they sound like anything at all), I ran a couple of percussive samples through a filter that produces a similar degree of group delay to a conventional two-way loudspeaker with a ported box. Discussion and files available here (you might need to read through a few posts on that thread to get a clear idea of what is being discussed and what is on the files).

No speaker with minimum phase crossovers can follow a square wave accurately. Faster attack is usually pointed to speakers with a bright signature which many speakerw with AMT and the likes show.

Transient response and perceived brightness are not related at all. AMT tweeters may have a faster transient response than other types of tweeters (or not - it depends on many factors, including the crossover), but if this is the case, then it's simply because their response extends up to a higher frequency.

Indeed it's essentially that simple: higher bandwidth = faster transient response.

Fortunately (except in extreme, implausible circumstances), this makes no audible difference.

 

[dfuller]

More often in my experience "speed" is more a function of cabinet design. Bass reflex systems have some "overhang" just by nature of how ports (and passive radiators, for that matter) work, which can sound like a speaker is "slower" but that's mostly in the bottom end and low mids, so tweeters aren't really part of the equation here.

 

[andreasmaaan]

More often in my experience "speed" is more a function of cabinet design. Bass reflex systems have some "overhang" just by nature of how ports (and passive radiators, for that matter) work, which can sound like a speaker is "slower" but that's mostly in the bottom end and low mids, so tweeters aren't really part of the equation here.

I suggest you try listening to files I posted the link to in my previous post^^

 

[dfuller]

I suggest you try listening to files I posted the link to in my previous post^^

Just did - the difference is subtle, but I did notice a difference. Sighted of course so I can't exactly depend on it, but the decay (so the speakers stopping) was noticeably different between 1 and 2 on both. Was it enough to spoil the listening experience (as it were)? No, not at all. Was it there? Yeah, a little.

 

[andreasmaaan]

Just did - the difference is subtle, but I did notice a difference. Sighted of course so I can't exactly depend on it, but the decay (so the speakers stopping) was noticeably different between 1 and 2 on both. Was it enough to spoil the listening experience (as it were)? No, not at all. Was it there? Yeah, a little.

Which version (1 or 2) had the decay in each case?

If you think you hear a difference sighted, then next step needs to be to test using something like Foobar ABX comparator

 

[dfuller]

Which version (1 or 2) had the decay in each case?

If you think you hear a difference sighted, then next step needs to be to test using something like Foobar ABX comparator

The longer decay was present on Revel 1 and Panotsi 2. Again, it was very subtle, and not something I'd notice if I wasn't doing a back-to-back A/B comparison.

 

[andreasmaaan]

The longer decay was present on Revel 1 and Panotsi 2. Again, it was very subtle, and not something I'd notice if I wasn't doing a back-to-back A/B comparison.

If you'd like to know whether you were correct, please pm me (but in any case, you need to do controlled trials for this to have any significance ofc)

 

[rkbates]

Not that long ago a member (sorry can't remember who) posted some graphs of measured transient response vs transient response calculated from measured frequency response. The graphs were extremely similar. The information captured in a transient response measurement is also captured in frequency and phase response. Frequency response is pretty easy to interpret, but I'm not sure to too many people can characterize slight differences in phase response or impulse response except to identify 'acceptable' or 'not-acceptable'. Nevertheless, I think phase response is under-rated when it comes to assessing system performance.