How to measure the loudspeakers actually work as they should?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I recently completed a 'Finalists' pair and I'm very happy with the result but I don't have a way to objectively judge them. Is it possible to measure and test them without special equipment?

And another question: people ask me how they can get the same sound from the available options in the market, the Finalists' design seem different than what I see in stores ("open" mid).
 
Looks a competent design:
Speaker Design Works

If you are fretting that everything is up to standard, you should check you have wired everything up right.

Face the speakers into each other faces and close up and listen to the bass. If you wire one speaker in reverse, the bass will cancel and disappear if those are right.

If the mids are wired wrong, the imaging of voices will sound disembodied and not central. Wrong tweeter phase or polarity is hard to spot. Physically check it. This speaker is wired +-+, so we might hope Jim Holtz knows what he is doing there.

The best test of a speaker is one of your favourite and familiar pieces of music. It should sound good on this three way, and go loud without effort.

Nothing very unusual about this speaker beyond it being a three way. It's a good way to do things. The specialist mid does voices and dispersion off-axis very well at low distortion levels.
 
If you have a another pair of speakers, or could borrow a pair, comparing and contrasting the 2 shows up any strengths and weaknesses much more obviously.

(If you can get a source of pink noise, that will show up any resonances. Also I find listening to spoken voices (particularly male) on the radio shows up colourations much more clearly than music.) But again, much easier to hear when comparing to other speakers.
 
Alan Shaw at Harbeth always reckons if a speaker can accurately reproduce his daughter's voice, he knows he is getting something right. :)

This is a fascinating explanation of building a very accurate loudspeaker: YouTube

The BBC had the advantage they could usually compare their monitor sound against live music too, by going into the other room.

I think you have done well to build this speaker. It is a big project, but should give years of pleasure. You could probably put loads of light BAF wadding in the open midrange tunnel to reduce colouration. But it has to be a good way of doing things.

In the long run, as you grow familiar with the whole speaker thing and even some modelling of your own, you might model a 3rd or 4th. order network on the tweeter. They improve distortion even more, IMO. But it's more effort and a popular kit maker tends to keep things simple.
 
Last edited:
If you have a another pair of speakers, or could borrow a pair, comparing and contrasting the 2 shows up any strengths and weaknesses much more obviously.

(If you can get a source of pink noise, that will show up any resonances. Also I find listening to spoken voices (particularly male) on the radio shows up colourations much more clearly than music.) But again, much easier to hear when comparing to other speakers.
That's a great advice, as system7 noted above our ears should be much better at telling when human voice sounds natural than music which goes through mixing.

Alan Shaw at Harbeth always reckons if a speaker can accurately reproduce his daughter's voice, he knows he is getting something right. :)

This is a fascinating explanation of building a very accurate loudspeaker: YouTube

The BBC had the advantage they could usually compare their monitor sound against live music too, by going into the other room.

I think you have done well to build this speaker. It is a big project, but should give years of pleasure. You could probably put loads of light BAF wadding in the open midrange tunnel to reduce colouration. But it has to be a good way of doing things.

In the long run, as you grow familiar with the whole speaker thing and even some modelling of your own, you might model a 3rd or 4th. order network on the tweeter. They improve distortion even more, IMO. But it's more effort and a popular kit maker tends to keep things simple.
I'm baffled... The more BAF the better? Currently there's only a third of the tunnel covered with foam.

Thanks for the link, I'm watching it right now (EDIT: it seems music isn't the best reference).
People asked me to build and sell them the same model, but this was too much work... next time I'll buy a used pair from a user hear and replace the old drivers with new ones... :D


Impedance curves gives the most help for the least effort and test gear.

For listening, a glide tone might be interesting.

B.
I just listened to a freq sweep, and there are many bumps and dips, every few herz's.

Looks a competent design:
Speaker Design Works
Thank you for the reply. My hearing isn't good enough to detect distortions and unequal response. I listened to a test recording and in the part where it played the same sound but in the wrong phase on the other channel I could still hear the sound normally. The response also seems to go down to around 50, past which the impact of the bass decreases fast (maybe I should cut from the foam to increase the breathing space?).

Again, it sounds really good to me, but I'm not used to listening to a decent set, and what's in objective terms is a distortion, to the ear used to the flawed set is euphonic.
 
Last edited:
We need pictures of this fine build. 3 ways sound initially dull, because they are so smooth and get so much right.

I think your ear gets trained to hear imaging and phasiness and dispersion and room response with experience.

And you need a well-treated (damped) room to hear everything on a recording, especially the microphone and studio acoustic. The room actually matters as much as the box.

This Finalists design is slightly vague on cabinet treatment. The BBC always did that well:
Interesting read I found on Lossy Cabinet designs by Harbeth

A bit of rubbery or felt panel damping stuck on goes a long way when a speaker is getting very good. Volume damping (fluffy stuff) depends on circumstances. Reflex doesn't want a lot near the port because the air needs to move. The mid in an open tube might benefit from a lot of wadding to control the organ-pipe type resonance near a quarter wavelength which here is going to be around 500Hz.

But all experimental. As is placement.
 
The glide sweep test does not lie, esp if it is a sweep-with-stop-watch glide. Not as helpful as REW, but I only mention it as a quick test needing no gear.

The questions are how to understand the gross results and then what to do about it for the coming year.

If things seem terrible, maybe the room is too bright like 95% of those pictured on this website. Does the sweep seem to have major octave-wide peaks or dips? Run out of steam much before your high-freq hearing does? And droop below 60 Hz, as it certainly will seem to if the speaker is performing for some wacky reason you've made it microphone-flat?*

B.
* "denial" of Fletcher-Munson effect is comparable to denial of vaccine effectiveness
 
Last edited:
^Those are "room modes" - interference of the sound waves bouncing between walls and floor-ceiling. Unavoidable, but you must find the best compromise for speakers and listening seat positions.

Room modes - Wikipedia

First you asked about measuring the speaker(s). My suggestion is to download the freeware Room EQWizard (REW), and use some USB microphone available (videocam?). Then follow instructions of REW, mic at 1m distance from a single speaker, record a sweep (computer's output to amp/speaker) and see how it looks. Then the other speaker too. For comparison, measure also some other speaker. Compare the result to what Jim Holz shows.

Actually REW is designed for measuring the reverberant sound in the room. You must make many measurements from each speaker to find out how room modes behave. If you focus to one frequency like 60Hz, set REW's sound generator to play 40/50/60Hz and walk around the room to hear how amplitude (strength) changes with modes. Then sit on your favourite position and as someone elso to move the speaker so that sound is fine. Etc. You can also use a SPL measuring application for a cell phone to support your ears.

Happy learning!
 
Last edited:
I don't have a way to objectively judge them. Is it possible to measure and test them without special equipment?

To do a true "objective" assessment, you do need some equipment, but it need not be expensive. You could do a lot with < $100 expense. Many here can tell you how to do that.

To determine what to actually measure you should look to Floyd Toole and his books and Sean Olive and his publications. They lead the field in understanding the very complex link between what we measure and how it sounds.

As Toole and Olive point out, personal listening is a technique that is highly fraught with danger and likely to mislead more than direct an endeavor for improvement.
 
Do you have a calibrated measurement mic, free REW, and a computer? If so, you can take some in-room measurements at the listening position to get a sense of the measured tonal balance.

There is good scientific research from Toole and Olive on correlating subjective listening preferences with in-room acoustic measurements, cross correlated back to anechoic speaker measurements. I wrote a summary here: The Science of Preferred Frequency Responses for Headphones and Loudspeakers.

At the end of the summary are links to several of Sean Olive's presentations which contain links to over a dozen AES papers on the topic. Sean peer reviewed my article for accuracy.

Putting room modes aside for a moment, a neutral sounding speaker will have a measured downward tilting response, at the listening position from 20 Hz to -10 dB @ 20 kHz. That's the findings from repeated subjective listening tests, correlated to objective in-room measurements from Harman's listening room. The test methodology they used is fascinating and in the links at the end of the summary above. As Toole says, accurate and preferred are synonymous.

Further, Tooles and Olive's research spawned a Standard Method of Measurement for In-Home Loudspeakers. That's a preview presentation as the standard costs about $100. I wrote a little summary about that, as I am really tired of bright sounding speakers and hoping speaker manufacturers adopt this new standard.

Anyway, hope that helps and good luck!

Mitch
 
At the end of the summary are links to several of Sean Olive's presentations which contain links to over a dozen AES papers on the topic. Sean peer reviewed my article for accuracy.

Putting room modes aside for a moment, a neutral sounding speaker will have a measured downward tilting response, at the listening position from 20 Hz to -10 dB @ 20 kHz. That's the findings from repeated subjective listening tests, correlated to objective in-room measurements from Harman's listening room. The test methodology they used is fascinating and in the links at the end of the summary above...
Thanks for links. Psychological testing is tricky.

If we had only that downward slope north of say 1kHz, it would hardly be noticed. On the other hand, HiFi's always sound bass deficient unless you boost the bass. Might be nothing but good old Fletcher-Munson.

Also, the Harman people showed that even experienced listeners can't tell bass boost from treble cut, not that any of us believe we would be confused.

So what I think is that systems never sound right without bass boost and only maybe with some lessening of the treble which cuts noise, sibilants, and other mic artifacts.

Basic problems are that there is no reference sound, no such thing as a "true" recording of Mahler in my living room, no reliable good memory, no adaptation level, no good language of sound....

B.
 
Last edited:
Hi Mitch

To me the downward slope comes about when we transition to more constant directivity (CD). CD puts a lot more sound into the room at HFs when compared to the older piston radiators which beam greatly. This means more sound in the reverb for the same direct sound - hence, CD sounds brighter.
 
Ben, with Tooles and Olives research, there is indeed a standard referenced in the links above. The subjective/objective controlled tests in Harman's "typical" listening room using hundreds of listening test subjects shows there is. The results are repeatable and the prediction of those results are embodied in the new ANSI/CTA-2034-A standard. Olives slides are really worthwhile as are the AES papers.

Note these target responses are calibrated at reference listening level in which are ears response are the most flat - around 83 dB SPL C weighting on average at the LP. As you know, our ears frequency response changes with changes in SPL. That's why I use JRiver's dynamic loudness control as I may not want to be listening at reference level. So even at 65 dB SPL, my subs still respond, and the sound remains full and balanced, but at a reduced SPL.

Hi Earl, one would think that, but in Toole's research, as long as the speaker exhibited good directivity control, whether lower or higher directivity did not factor into the neutral preference. I have tried it on cones and domes and constant directivity speakers and the same target applies with no changes. The tonal response sounds the same to my ears. But the stereo presentation sounds different to my ears. Cones and domes more distant or diffuse, whereas waveguides more direct is how I would subjectively characterise it.

Sorry lamir35 if this is too much OT.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.