Understanding Speaker Measurements (Video)

[312elements]

Thank you for the video. It was very informative. Could someone explain why one wouldn't use the REW filters to correct for the response instead of an informed (by measurements) educated guess. Then take that once step further, and confirm with measurements that the adjustments solved the problem and didn't create any additional problems? I could see not wanting to boost frequencies and increase distortion in the process, but if we were to just focus on cutting frequencies, wouldn't the computed filters with a confirmed measurement be a more accurate way of correcting the problem?

 

[pjn]

Per popular demand, I recorded a long (1 hour) video on how to read speaker measurements but also, the underlying research (and physics) that mandate some of those concepts. Also covered is equalization and listening tests (briefly). As usual, my videos are "one take" directly recorded and uploaded (and in this case, with no reharsal). So please take it easy as far as critiquing it in that regard.

A personal note: as you all know, I don't allow advertising in my videos, nor do I pester people to subscribe, give likes, etc. in the video. Alas, it seems some out there are keeping score on popularity of my content as proof point of whether their audio videos are more correct than mine. So let's even the playground and subscribe to my channel if you have not before. Click on this link to go to my youtube channel (youtube.com/@audiosciencereview) and do that. Appreciate it in advance.

Great overview - thanks!

 

[JktHifi]

There is a portable Dayton measurement mic and an app called Audiotools that can do similar stuff like REW.
But I don’t think most manufacturers would allow you to play sine sweeps and pink noise to expose their bad speaker designs

It’s very nice to have this. Thanks

 

[Oristo]

Could someone explain why one wouldn't use the REW filters to correct for the response instead of an informed (by measurements) educated guess.

One can use REW; I and many do. Considering Amir's explanation about on and off-axis responses (IME among the best), then measurement technique becomes crucial.
Specifically, for a speaker with strongly mismatched on and off-axis response in untreated (lots of echo) rooms, measurements using a highly directional microphone will not be representative of what you hear. (Perhaps by dumb luck, my stereo speakers seem better than average for on vs off-axis response.)

FWIW, concerns with first third of the video:
1) implication that long wavelengths will be omnidirectional IN ROOMS; bass standing wave peaks and nulls confound that.
2) Amir's chuckles may seem condescending and triggering to viewers for which the content is unobvious or contrary to their prejudices.

 

[Jeromeof]

Amir doesn't exactly say step response is useless. He says such a signal doesn't exist in real World and the dynamics of the drivers to recover (in your words) are irrelevant. However, he mentioned the time delay between the tweeter and midbass driver the step response illustrates. Minimizing this delay in speaker design is important for heightened sound stage. Some high-end speakers even come with adjustable tweeter, midbass angles and lots of instructions to set it up correctly for the actual LP distance.

Thanks but I totally get the different time delays and importance of the distance between the tweeter and woofer, I am seriously thinking of getting a Kef speaker as my next speaker purchase for this reason.

But that is a different delay, my question is about the physical recovery time for a single driver when asked to play 2 separate tones milliseconds apart, once a crossover / high low pass filter is involved this isn't a problem as one tone would goto the tweeter and the next one woofer etc but when both tones are 'almost on top of each other' in time (i.e. milliseconds) and both need to be played by the same driver

 

[VQR]

I forgot that (extremely low) trolling quality in YT comments...
View attachment 281132

View attachment 281133
View attachment 281134

You can tell neither have read a single ASR review of a speaker in their life.

 

[Cote Dazur]

Thank you @amirm, a stellar presentation on a very dry subject.
At least now, after seeing the video (I might have to go trough it a few more time) those graphs are less of a hieroglyph and more of a lesser know language with recognizable calligraphy to me.
At least now, when I read "resonance" I have a much clearer picture of what it looks like and why it may or may not matter.
Same for "directivity"
Would love to see the anechoic frequency responses per CEA-2034/ANSI measurement for my speakers, but it probably won't happen just because of their weight and size. Lucky for me, I have dedicated rooms where they can be positioned to sound good with enough time and patience.
As a side note, interesting to hear you say the front wall is the wall in front of the speaker face, as for me I always though of it from the perspective of the listener. Is there not a standard way about what the front and the back walls are?
Keep up the good work, explaining what you are doing and why goes a long way for many, as just looking at the graphs without the lexicon is a little useless.

 

[Oristo]

As mentioned in Amir's video, much work on [mis]perception of loudspeaker quality by listeners was formalized by Dr. Floyd Toole,
with his most frequently cited papers still behind AES paywall. However, a preprint summarizing some of that is available here:
The Acoustics and Psychoacoustics of Loudspeakers and Rooms-The Stereo Past and the Multichannel Future
Included is this graph:

... commenting: "Listeners with hearing loss not only exhibit high judgment variability, they can also exhibit strong individualistic biases in their judgments."

 

[Timcognito]

Missed this yesterday watching the GS Warriors barely win verus the Sacramento Kings. This will be tonights entertainment. Thanks @amirm for all the effort. Like others I've been hoping for this.
Shout out to @AdamG247 for help with a problem that I thought was the site's but really on my end. Glad that he was there to help on this fun and informative site.

 

[jomo]

The volume of my room is very large (open space with 25+ feet ceilings) so my room modes are not severe. But they are surely there and would need correction for best response depending on the speaker. For my main system where I test consumer speakers, I routinely dial a PEQ at 105 Hz for this purpose (I note it in the reviews). I don't usually bother for near-field.

Net, net, EQ for bass is mandatory in all situations. Even rooms that are heavily treated still have modes.

Thanks Amir for the tutorial. From what I gather, the Klippel calculated reflection responses published in ASR are related to your home situation, not necessarily generic. Being that some of us live in homes that are acoustically a mess and are difficult to treat without issue, is there Klippel computational information that can be helpful with different room situations? For example, the effects of the dimensions/orientation of the room, speaker position in the room, wall materials (drywall, plaster, glass,...), ceiling height and shape (flat, coved...). Then from that information, the best means to attack the room acoustic problems (acoustic tubes, wall panels, ...). I've listened to very accurate speakers in my home, and the room dominated the sound.

 

[312elements]

@jomo the way i understand it there are two major wars we fight in terms of in room response of two speakers. Decay time and frequency response. To an extent, frequency response can be beaten into submission, at least the peaks with DSP. I don’t believe there is a digital cure for RT60. Do you have a umik? I realize that this isn’t what you’re asking and I apologize in advance if this sounds forward, but even understanding measurements of other speakers has little value in your own space without measurements.

 

[jomo]

@jomo the way i understand it there are two major wars we fight in terms of in room response of two speakers. Decay time and frequency response. To an extent, frequency response can be beaten into submission, at least the peaks with DSP. I don’t believe there is a digital cure for RT60. Do you have a umik? I realize that this isn’t what you’re asking and I apologize in advance if this sounds forward, but even understanding measurements of other speakers has little value in your own space without measurements.

I do have a umik and have used it to run a few REW frequency responses. But that's about it. I don't know how to use the information to correct my room.

 

[ROOSKIE]

Thanks Amir for the tutorial. From what I gather, the Klippel calculated reflection responses published in ASR are related to your home situation, not necessarily generic. Being that some of us live in homes that are acoustically a mess and are difficult to treat without issue, is there Klippel computational information that can be helpful with different room situations? For example, the effects of the dimensions/orientation of the room, speaker position in the room, wall materials (drywall, plaster, glass,...), ceiling height and shape (flat, coved...). Then from that information, the best means to attack the room acoustic problems (acoustic tubes, wall panels, ...). I've listened to very accurate speakers in my home, and the room dominated the sound.

No, there is a misunderstanding here. The Klipple in room 1st reflections are generic predictions.
They have zero to do with Amir's listening room.

 

[Oristo]

For an appreciation of how well appropriate objective test results correlate with double-blind listening tests, consider the Full video interview of Dr. Toole, starting around 29:00 / 41:46 https://www.namm.org/library/oral-history/floyd-toole "There is no excuse for bad sound anymore". He tells of poor "Consumer Reports" ratings for Harmon products motivating that research report for appropriate measurements to predict double-blind preference test results.

 

[ROOSKIE]

I do have a umik and have used it to run a few REW frequency responses. But that's about it. I don't know how to use the information to correct my room.

Based on room size you have a transition zone where the speaker is handing off much of the acoustic experience to the room itself.

usually this is from 200-400hrz again depending on the room.

• Below this I 'correct/adapt' the sound strongly via DSP to account for the room issues. Though I do not micromanage the response, my goal is smoothish below transition not perfectly smooth. (and definetly not flat when in room, flat is for anechoic data, not in a typical room which usually slopes down from bass and usually will sound best with some bass boost, per personal preferences )
• In the zone of 200-400hrz I sometimes adjust and sometimes I do not depending on the exact issue and experimentation.
• Above the 200-400hrz zone I usually only correct the speaker based on the Klipple testing data and not much on in room data.

Fully implementing smart speaker and room correction/adaptation is a complex endevor. Plan to learn and spend some time though it is worth it. Once you have basically good gear no other attempt at improvement will likley ever compare. Also bear in mind there are a few diffent approaches and many folks discussing what is what so there will be some desicions to be made amidst some controversy.

 

[Oristo]

I do have a umik and have used it to run a few REW frequency responses. But that's about it. I don't know how to use the information to correct my room.

Unless your microphone has a sensitivity pattern that approximates paired HRTFs, wide-band room correction is unrealistic.
However, as Dr. Toole points out, about 1/3 of perceived audio quality is bass, and as Amir points out, bass is less problematic to measure
(being less directional), and problematic in real rooms thanks to standing waves. Consequently, by employing speakers with good ASR tests results,
applying room treatment at first reflection points and using REW to minimize bass peaks and nulls (preferably first by subwoofer positioning), you should be good.

 

[amirm]

From what I gather, the Klippel calculated reflection responses published in ASR are related to your home situation, not necessarily generic.

I thought I explained this already.

Klippel's core functionality is just to measure the soundfield. It then has an optional post processing module to generate frequency responses per CEA/CTA-2034/ANSI standard. It is those specifications which mandate how to generate those graphs including early reflections. That standard is based on research at Harman where they analyzed the reflections in a number of different listening spaces and came up with a typical set to use.

Here is the paper if you want to read more:
Characterizing the Amplitude Response of Loudspeaker Systems
Allan Devantier, Harman International Industries Inc.,

"The amplitude response of a loudspeaker system is characterized by a series of spatially averaged
measurements. The proposed approach recognizes that the listener hears three acoustical events in a typical
domestic environment: the direct sound, the early arrivals and the reverberant sound field. A survey of 15
domestic multi-channel installations was used to determine the typical angle of the direct sound and the early
arrivals. The reflected sound that arrives at the listener after encountering only one room boundary is used to
approximate the early arrivals, and the total sound power is used to approximate the reverberant sound field.
Two unique directivity indices are also defined and the in-room response of the loudspeaker is predicted from
anechoic data."

Ends with:

SUMMARY AND CONCLUSIONS
We have developed a concise set of amplitude
response measurements that are based on the three
acoustic events heard in a typical domestic
environment: the direct sound, the early reflections,
and the reverberant sound field. The geometry of
fifteen domestic listening environments was analyzed
to determine the angle and attenuation of the direct
sound and the early reflections. It has been
demonstrated that the early reflections are strong
enough in a typical listening environment to affect
the sound and spatial quality of the loudspeaker –
room system. We have also shown that gated
loudspeaker measurements might not have enough
frequency resolution to reliably identify potentially
audible resonances in the critical midrange. Finally,
we are able to accurately predict the in-room
response of a loudspeaker in the critical midrange
region.

We have shown that these amplitude response
measurements correlate with listener preferences in
well-controlled double blind listening tests. For a
loudspeaker to perform well in these tests the direct
sound, the early reflections and the reverberant
sound field must all be considered.

In other words, data from the 15 prototype rooms were shown to average out to something with predictive power in controlled listening tests elsewhere.

 

[amirm]

I do have a umik and have used it to run a few REW frequency responses. But that's about it. I don't know how to use the information to correct my room.

I will need to do a video on this. For now, you want to measure your room using REW and a mic. Then look at the variations in bass frequencies. To the extent there are peaks, you can apply filters with negative gain to pull them down. You also need to then shape the overall response across full audible range to have a slope from high to low. These two things are mandatory to get proper sound regardless of which speaker you buy, and what room you put it in.

 

[Eytsch]

I have not but the results would be as you would predict. Unless the speaker is compressing or something, the DSP does exactly what you would expect.

"speaker is compressing" is a suspicious line that gets used frequently in guitar amp discussions online. I'd love to see some measurements (or solid arguments) proving or disproving it.

 

[thecheapseats]

...I thought I explained this already....

you did - it was metioned explicitly in the vid tutorial - that reflections were approximations - based upon the Klippel data plugged in to an algo that yielded appx reflections...