Frequency response is everything!?...

[Ron Resnick]

Yes I use DSP and DRC to sew the system together seamlessly.

If you didn't use any DSP or DRC, what do you think your frequency curve would look like?

 

[Keith_W]

If you didn't use any DSP or DRC, what do you think your frequency curve would look like?

I can show you mine. The speakers are Acapella High Violons.



This curve was taken in 2010 or thereabouts. The passive crossover was in place, and the speaker was driven by a single amplifier. Measurement was taken at the listening position (same room as currently). Some comments:

- Bass fall-off below 50Hz at about 30Hz/octave.
- Wide Q suckout centered around 200Hz, probably a room mode.
- Rising frequency response after 6kHz from the tweeter (crossed over at 5kHz).

Listening impressions: Overall a lean sounding speaker with an amazing top end and a woolly bottom end. Not enough bass. I bought these speakers after listening to them, knowing that I would have to do something to fix the bass. After a lot of dead ends, I turned to DSP.



This is the current system, before correction. This was taken 4 days ago. Same speakers, same room, but the passive crossover has been bypassed. Each driver is connected directly to its own power amp channel and its own DAC channel (I have 8 DAC/power amp channels in total, for 2 subs, 2 woofers, 2 midrange horns, and 2 tweeters). Notes:

- IGNORE the rising response in the treble above 7-8kHz. I suspect my microphone has developed some kind of issue. They certainly don't sound like that! I decided not to correct the high frequencies and leave them alone.
- Note that the scale is different so they are not directly visually comparable. The first graph is +20dB to -90dB (110dB vertical scale). The second graph is +40dB to -10dB (50dB vertical scale). If you were to correct for the scale and superimpose the graphs, then every peak and dip in the first graph would be twice as large.
- You can "almost" see the crossover points, which are 80Hz, 500Hz, and 5000Hz. This is because there are different amps driving them, all with different input sensitivities and output gain. I typically do a sweep like this, and then adjust the trim of each pair of amplifiers until it is as flat as possible, repeating sweeps to confirm as I go along.
- You can see the influence of the room, particularly around 50Hz which is probably a room mode.

Listening impressions: once all the amplifiers are level matched, with only the digital crossovers and no correction, the system sounds disconnected. This is caused by group delay (each driver has group delay with respect to the other drivers). Bass inconsistency is obvious and manifests by some loud bass notes and some which are too soft.



This is after matching the trim of all the amplifiers, correcting the drivers (I decided to leave the midrange horn and tweeter alone), time aligning all the drivers, and applying overall room correction. Notes:

- The rising treble is still there (again, faulty microphone!)
- All the peaks are gone, and what remains are only the dips which I have deliberately not corrected. The peaks are +/- 2dB at the listening position, which is superb performance, especially in the bass.
- You might note that I deliberately rolled off the bass from 30Hz. This is because furniture and the walls rattle at about 25Hz. In the past, I have tried to fix the rattling but I now think it's the plumbing or something inside the walls. I can't do anything about that (short of tearing down the wall) so a quick and dirty fix is to simply remove the problem frequency.
- The falling response is from my own target curve: 0dB at 20Hz, -4dB at 1kHz, and -8dB at 21kHz. I did not expect the massive fall-off above 5kHz (which is where I crossed over the tweeter). It turned out to be an error in the target curve design. I have not yet gone back to fix it.

Listening: the bass issues are fixed. Any bass notes sound even with no loud notes. Transients have real impact. Still not happy with my target curve - a flat frequency response at the listening position always sounds bright, and I did not apply the Harman target. I have the ability to play with the individual volume trim of all the amplifiers (works like a rough EQ) and actually apply digital EQ. I have taken note of these settings and I will design my next target curve around it.

When I have time, I will obtain a new microphone, redo all the measurements, and apply my own target curve (not the same as the Harman target). I don't know what it will sound like yet, but it will hopefully be an improvement.

 

[henrich3]

First of all allow me to reveal my bias towards him. He banned me from ASR.
IMO he does not put any stock in subjective listening test
It just civer for his all to frequent hatchet jobs on products he feels do not follow his technical prejudice
He did a similar hatchet job on Marton Logan.

I now understand the basis for your ad hominem attacks on amirm. I don't think that we need to discuss your views on him or ASR any further.

 

[henrich3]

In the particular example you gave I wonder if the kippell system can measure a dipole in the nearfield as you need distance to resolve the rear wave cancelation and supporr frequencies ... not doubting they measure poorly but there would need to be very specific input and smart math to predict a farfield response .. in that case the room takes over anyhow.
Linkwitz measured outside and up high to remove reflections.
I recall magneplain never submitted for reviews that measured
Cheers
Phil

Some of the measurements that amirm posted for that speaker go over my head, like the Klippel Apparent Sound Power measurement that you referenced. I appreciate the thoroughness that he went to with the measurements, but I'm not well educated enough to know how to interpret some of them. They do, however, give me greater confidence in his conclusions about the product. Reading a typical sighted subjectivist equipment review that contains nothing more than flowery prose just doesn't inspire the same degree of confidence. I like measurements with my flowery prose...

 

[Gregadd]

I now understand the basis for your ad hominem attacks on amirm. I don't think that we need to discuss your views on him or ASR any further.

Ad hominem? Not at all. I regarded him as friend.
It is he who went over the deep end. He also blocked me from his you tube channel. I am ready to discuss audio with him anytime.

BTW it is you who brought him up.

 

[Keith_W]

Some of the measurements that amirm posted for that speaker go over my head, like the Klippel Apparent Sound Power measurement that you referenced. I appreciate the thoroughness that he went to with the measurements, but I'm not well educated enough to know how to interpret some of them. They do, however, give me greater confidence in his conclusions about the product. Reading a typical sighted subjectivist equipment review that contains nothing more than flowery prose just doesn't inspire the same degree of confidence. I like measurements with my flowery prose...

Measurements are objective. But interpretation of measurements is subjective and just as prone to human error. For example, the measurement may be incorrect (see my earlier post about the pitfalls of measuring headphones). Or too much weight might be given on a single metric. Or assumptions made about how equipment might behave in another state when the measurement was made at a steady state. Or being unable to perform the actual measurement (e.g. dynamics or soundstage) and inferring performance from indirect metrics.

None of this means we should abandon this endeavour altogether. Running measurements is a great tool, but like all tools, tools are stupid. It is the wielder of the tool who is intelligent (at least, you would hope so).

Understanding the meaning the Spinorama (of which that Klippel apparent sound power measurement is part of) is quite easy. If you don't understand it, I suggest reading this or this.

 

[ecwl]

I’ve been refraining from commenting.
I think frequency response definitely is not everything. As others have mentioned, time domain issues matter. And there are lots of easily measureable things that can affect sound like your impulse/step response, RT60/resonance/reverb, harmonic distortions.
I also think there are also many less frequently measured things that I believe affect sound quality like noise floor modulation or jitter.
And then there are things that are not measureable that seems to affect sound quality like low level noise that are not measurable which seems to indicate cable differences in some instances are real. Rob Watts the Chord DAC designer has noticed that his DAC’s digital accuracy at very low levels, beyond known audibility (-120 to -300dB) has audible effects that he doesn’t fully understand and perhaps that explains why cable differences matter.
but with all that said, I also think that it’s not possible to have a truly great sound system or sonic experience if the frequency response is significantly out of whack.
On the other hand, stereos are hobbies and about personal enjoyment. I’ve seen many people who prefer a bespoke frequency response that I don’t like or extra euphoric harmonic distortion that’s not my cup of tea. And really, if they love what they hear, who am I to judge.

 

[schlager]

If you didn't use any DSP or DRC, what do you think your frequency curve would look like?

Worse a normal room screw up any FR in SS, no matter what speaker you use. Most compression driver tweeters needs EQ because of the mass roll off, in the bass we all know what problems we are fighting. Very uneven FR below the transition frequency is the norm and I can't stand it, especially when you are used to an nice and even in-room FR. As long one know what to correct, then after EQ, the sound quality is always improved considerably.

 

[Hear Here]

after EQ, the sound quality is always improved considerably.

Only in the bass and only if the top end completely bypasses any singnal processor - ie you need a bi-amping system with DSP only appled in the bass amp.

DSP built into a full-range amp may well flatten the frequency curve (particularly if the owner can't or won't do this in less invasive ways), but it will reduce the sparkle and goosebump factor that un-proceesed top end can and should provide.

 

[Ron Resnick]

Worse a normal room screw up any FR in SS, no matter what speaker you use. Most compression driver tweeters needs EQ because of the mass roll off, in the bass we all know what problems we are fighting. Very uneven FR below the transition frequency is the norm and I can't stand it, especially when you are used to an nice and even in-room FR. As long one know what to correct, then after EQ, the sound quality is always improved considerably.

Thank you.

 

[schlager]

Can't measure soundstage depth or width. Can't measure dynamics (and no, it's not the step response)

Soundstage depth or width, in short is related to direct vs reflected sound, speaker directivity, off-axis FR. Different mix of direct vs reflected sound, will give another subjectively presentation of soundstage, depth and width.

Speakers with controlled directivity to under 300 Hz or so, gives more clarity and a better defined physical and separated presentation. Even off-axis FR gives a better imaging, while a wider sound pattern (most speakers), gives a wider panorama and spatial feeling.

For dynamics it gets more complicated, but it can be translated into measurements and no step response would be true time alignment aka distance and phase.

 

[schlager]

Thank you.

Now you can sleep tight again, LOL.

 

[hopkins]

For dynamics it gets more complicated, but it can be translated into measurements and no step response would be true time alignment aka distance and phase.

Can you elaborate ?
Thanks

 

[schlager]

Can you elaborate ?

Our hearing mechanism interprets dynamics more as transients response, or that would be the hifi systems (mostly speakers) ability to reproduce transients. REW and a mic will show it on a decay plot, the better system will have a cleaner, faster early decay. You would also be able to see it on distortion and compression analysis. A big speaker with high efficiency, will sound more realistic and dynamic, compared to a small 2-way standmount, even if they had the same FR, due to less distortion and compression. Let me point out that dynamics has nothing to do with SPL per se.
I would also say that speaker directivity comes into play, because it affects what happens to the signal in time, due to how the reflected sound adds to the original signal. If there is less early reflections, the transient will be more precise and the leading edge of the signal (musical notes) will start and stop more quickly with less smearing of the sound. I'm sure there is (a lot) more to the subject, but that is my initial thoughts.

 

[Keith_W]

Soundstage depth or width, in short is related to direct vs reflected sound, speaker directivity, off-axis FR. Different mix of direct vs reflected sound, will give another subjectively presentation of soundstage, depth and width.

Speakers with controlled directivity to under 300 Hz or so, gives more clarity and a better defined physical and separated presentation. Even off-axis FR gives a better imaging, while a wider sound pattern (most speakers), gives a wider panorama and spatial feeling.

I know that, hence I made my point about inferring performance from indirect measurement. A wide, smooth radiation pattern is more likely to give you a wide soundstage and depth, however the speaker is only half of the equation. The other unknown is the interaction between the speaker and the room. Any reflected sound arriving at your ears above 40ms will be perceived as an echo or reverb (Haas effect). The spectral content, amplitude, delay, decay, and direction of the reflection is important. The Spinorama ONLY helps you predict the spectral content and none of the others. Sure, you can measure reflections with a microphone, but you can not determine the direction of the reflections - at least without going through a lot of trouble. My mic is an omni (Earthworks M30) and it might be possible to measure directions if I put some waveguide around it and pointed it at different directions? I have never seen or heard of anybody doing this however.

For dynamics it gets more complicated, but it can be translated into measurements and no step response would be true time alignment aka distance and phase.

I was wondering how to measure dynamics hence I started a thread on ASR asking this question. Perhaps it might be a reflection of the people who saw the question and answered it, but my take-away from that thread is that there is no way to measure it? I thought that was strange, because it is a physical property that should be measurable.

 

[henrich3]

I know that, hence I made my point about inferring performance from indirect measurement. A wide, smooth radiation pattern is more likely to give you a wide soundstage and depth, however the speaker is only half of the equation. The other unknown is the interaction between the speaker and the room. Any reflected sound arriving at your ears above 40ms will be perceived as an echo or reverb (Haas effect). The spectral content, amplitude, delay, decay, and direction of the reflection is important. The Spinorama ONLY helps you predict the spectral content and none of the others. Sure, you can measure reflections with a microphone, but you can not determine the direction of the reflections - at least without going through a lot of trouble. My mic is an omni (Earthworks M30) and it might be possible to measure directions if I put some waveguide around it and pointed it at different directions? I have never seen or heard of anybody doing this however.

I was wondering how to measure dynamics hence I started a thread on ASR asking this question. Perhaps it might be a reflection of the people who saw the question and answered it, but my take-away from that thread is that there is no way to measure it? I thought that was strange, because it is a physical property that should be measurable.

The mic used to setup a Trinnov Altitude pre/pro can tell the direction of a sound -

https://www.trinnov.com/en/products/altitude-sup-16-sup/

"During calibration, Trinnov’s unique 3D microphone precisely maps the location of each loudspeaker in the room regarding distance, azimuth, and elevation."

 

[hopkins]

Our hearing mechanism interprets dynamics more as transients response, or that would be the hifi systems (mostly speakers) ability to reproduce transients. REW and a mic will show it on a decay plot, the better system will have a cleaner, faster early decay. You would also be able to see it on distortion and compression analysis. A big speaker with high efficiency, will sound more realistic and dynamic, compared to a small 2-way standmount, even if they had the same FR, due to less distortion and compression. Let me point out that dynamics has nothing to do with SPL per se.
I would also say that speaker directivity comes into play, because it affects what happens to the signal in time, due to how the reflected sound adds to the original signal. If there is less early reflections, the transient will be more precise and the leading edge of the signal (musical notes) will start and stop more quickly with less smearing of the sound. I'm sure there is (a lot) more to the subject, but that is my initial thoughts.

Thanks.
From a layman's perspective, what measurements will not capture is the ability of a speaker to reproduce a more complex signal (real music) with little distortion. How do you evaluate with a frequency sweep the ability of a driver to handle quasi simultaneously a wide range of frequencies?

This is not only relevant for single driver speakers, by the way (though even more critical for those, obviously). For single driver speakers, measurements are even less relevant in this aspect, which is, however, critical to musical enjoyment IMO.

 

[schlager]

How do you evaluate with a frequency sweep the ability of a driver to handle quasi simultaneously a wide range of frequencies?

You can't with a frequency sweep, but pink noise and RTA measurement would probably be the way to go. Squarewaves could also be used, I recon. But how is this related to dynamics?

I find it logical that speaker efficiency comes into play. Horns and line arrays are normally perceived as dynamic speakers, they typically have low distortion and compression.

About how we measure it, I still stand by my reasoning in post #194. Of course there could be other ways to do it, but I haven't figured it out (yet). Someone grab the ball.

 

[hopkins]

You can't with a frequency sweep, but pink noise and RTA measurement would probably be the way to go. Squarewaves could also be used, I recon. But how is this related to dynamics?

I find it logical that speaker efficiency comes into play. Horns and line arrays are normally perceived as dynamic speakers, they typically have low distortion and compression.

About how we measure it, I still stand by my reasoning in post #194. Of course there could be other ways to do it, but I haven't figured it out (yet). Someone grab the ball.

Thanks. It was my assumption that the ability of a speaker to produce accurately multiple frequencies also had an impact on "dynamics" - it would seem logical that anything impacting the accuracy of a speaker cone's movement would translate in terms of dynamics - but I could be wrong. Anyway, it is clear that SPL is a useful but not sufficient indicator for speaker performance, as many have already stated.

 

[PeterA]

I remain amazed when I think that a good two way speaker can present the many instruments of an orchestra all playing together with the large range in timbres and volumes, and still sound convincing. It's really an amazing technology when you think about it.