Should we (I) get into speaker testing & measurement

[Sancus]

I voted yes and in addition state that I would contribute at least a few hundred dollars to the cause should you begin funding this.

Finding speaker measurements is far too difficult and addressing that is probably the greatest challenge this site could take on. That's a good thing. But I do understand that it may prove logistically and financially impractical.

 

[Blumlein 88]

I have a question about the loudspeaker measurement method here:
What about dynamics?

What's important for loudspeaker subjective quality is not just the frequency response and directivity.
It's also the distortion and the ability to transcript dynamics.
How is this method doing in that regard?

Another point of concern: there may be 'good enough' DAC at $99.
You've been able, through your measurements, to create a credible definition of 'good enough', in that case.

But will you be able to have such a definition of 'good enough' for loudspeaker that will work for the majority of us ?
As for frequency response and directivity, I guess, Toole's work is our bible
But then will you be able to find any 'cheap' loudspeakers falling in that category?
How objective and measurable will that be, if you have to choose between several shortcomings?
And are the frequency response and directivity enough to define a 'good enough' speaker?

Soundstage.net does some measures where they compare the response of the speaker at 70 db and 95 db SPL at 2 meters. You are comparing the curves to see how they differ. If there is some change with a change in level. For the most part those are straight lines. So I don't think changes with dynamic levels is much of an issue.

One thing Klippel has done work on is cone rub, or cone buzz or other extraneous sounds when speakers are moving. I don't know how audible that is or what the level of knowledge about them are, but it seems interesting.

As for cheap speakers that work somewhere around good enough, I think the JBL LSR305 is the poster child for that. They are surprisingly good, designed by Harman, and $150 each. Oh and that includes the amps as they are self powered. Separate amps for woofer and tweeter.

 

[soundwave76]

Yes, even though it will probably be a difficult road. How about discussing some sort of a joint partnership/venture with the measurement company, where both would benefit and that would bring the cost down?

As an active DSP monitor fanboy myself (Genelec/Kii etc) I see the future of speaker sizes leaning towards small-medium sized that can be shipped relativelly cheap.

 

[Krunok]

What the Spinorama doesn’t show is distortion and compression, which the NRC’s measurements (on SoundStageNetwork’s site) do show, which is also very helpful, as usual speaker measurements are done simply at 2.83V. The Spinorama can be amazing, but if it’s an 85dB sensitive speaker that distorts when fed 10W and up, then it’s only suited for desks and bedrooms.

+1 for distortion and compression. Can this device measure them?

 

[Krunok]

One thing Klippel has done work on is cone rub, or cone buzz or other extraneous sounds when speakers are moving. I don't know how audible that is or what the level of knowledge about them are, but it seems interesting.

Wouldn't cone rub result in a distortion?

Btw, what Soundstage doesn't measure is 105dB (or 100) response and distortion, which i would be very interested to see.

 

[Blumlein 88]

https://www.klippel.de/fileadmin/_migrated/content_uploads/Loudspeaker_Nonlinearities–Causes_Parameters_Symptoms_01.pdf

Its been some time since I read through it. But this is one of the better treatments of non-linearities in loudspeakers. 60+ pages. Packed with good information.

You also might try one of their up/down listening tests online.
http://www.klippel.de/listeningtest/lt/

 

[KSTR]

I voted no. Why? I've done speaker measurements in my day job. It's not fun at all, it's damn hard time consuming work and the investment into equipment and procedures is huge if you want to do it efficiently on a regular basis (you don't need a Klippel at all, but you will want a computer controlled robot arm for usable 3D directivity data, http://fouraudio.com/en/products/elf.html). If you don't have a half-decent large anechoic room (one single reflective surface, the floor, where the mic is mounted into) you'll have extra work with properly windowing the IR's and combining those with nearfield data, which is an art by itself (both the windowing and the combining).
Mind you, to characterize a speaker, you may need on the order of 1,000 individual measurements and you need to present them as well (no way just showing some plots alone, you'd need to provide the actual IR data, preferably in an industry standard format).

As much as I would applaud any effort to make ASR a speaker measurements database as well, I strongly advise against it unless you are fully aware what it takes to create reliable and repeatable data, even when no absolute accuracy is aimed for. You need funds and you need manpower to do that.

 

[Wombat]

Common-sense re cost and effort gives a NO! for a one-man band.

 

[amirm]

Keep the comments coming guys. You are giving me a number of ideas, thoughts to chew on.

 

[amirm]

It would be the natural next step for ASR, i also believe most of what we look for actually comes from the speaker. Getting big ones might be difficult but i think regular, smaller, more affordable speakers would be as interesting to review, the hundreds of bookshelves in the 250/500/1000$ range currently available is a lot a material to work with already. Not even counting all the studio/active monitors part of the market, which im sure many professionals would even pay to get this kind of datas.

That is my current thought. ASR's path on reviewing electronics started with a casual question of, "what are some under $100 DACs that are worthwhile?" We covered that, generated a lot of excited for really not much money, and we moved forward.

I am shocked at how cheap bookshelf speakers, and monitors are these days. I can imagine members purchasing these, drop shipping them to me for testing and then shipping forward. Yes, the shipping cost will be higher than electronics but don't we need to take some steps to bring order to this madness? Surely there are a lot of terrible and good products waiting for us to discover.

Through Madrona, I should be able to get access to a few speakers.

Also, I have this hope that the local audiophiles here will be willing to bring their speakers here for testing. This is why a fast, automated system like Klippel is worthwhile. High resolution measurements are made all automated with just a bit of setup. The schemes using Quasi-anechoic as is done elsewhere is just too time consuming and too manual. I know I don't have the time, patience or interest in doing that type of work.

I can easily see access to 20 to 30 speakers to test to get things going. So right now, I am not worried about supply of the speakers. Shipping cost, handling the weight, etc. remains a small concern but not as big of a hurdle relative to benefit.

 

[amirm]

@amirm Can you turn it all into an academic study, and apply for a grant for the test equipment?

Not a grant or anything like you are talking about but yes, I am looking at reducing the net cost. But even then, we are talking significant dollars that needs to be funded.

 

[amirm]

If measuring speakers is too challenging financially or logistically, perhaps headphone testing might be a better bang for the buck, since people naturally come here for headphone amp and dac reviews. I would love to see headphone measurements compared to the Harman target model and ranked.

I am concurrently researching headphone testing with this project. It has its own set of issues. Cost for a high-end system is lower but we are still talking $10K to $20K. These guys charge a fortune for everything.

I will probably start a new thread on headphone testing and get feedback on interest and issues involved. From logistics point of view, headphones are of course much easier to manage. The Klippel system at the extreme requires 14 foot high ceilings! I only have one spot in the entire house for it and is definitely not ideal.

 

[amirm]

If you don't have a half-decent large anechoic room (one single reflective surface, the floor, where the mic is mounted into) you'll have extra work with properly windowing the IR's and combining those with nearfield data, which is an art by itself (both the windowing and the combining).

The Klippel system solves this problem. It makes measurements at two distances from the source and uses that to compute the reflections and subtracts them out.

The dashed red line is the measurement as polluted by reflections. The green are the reflections. The solid blue is the extracted, correct response.

The nice thing about this system is that the speaker stays stationary and the lightweight microphone moves around. This allows measurement of any weight or size speaker. Robotic arms that swing the speaker around have severe load limitations.

 

[amirm]

The Benjamin paper had a nice overview (high pass the stimulus to shorten impulse length, reverse it in processing, use lots of averaging to get SNR back) but unfortunately I loaned out my hard copy.

I've attached a bit of an evolution on this technique, using a notch. It includes good references, Fincham @ KEF thought of it first.

Thanks for the references. Unfortunately all of these schemes require work, manual intervention to get reasonable results. Add this to manual measurements at multiple locations and you have a long process for measurements.

The "innovation" I am suggesting here through the use of Klippel system is what makes this whole thing work: a fast measurement system that doesn't require anechoic chamber, produces processed graphs compliant with standard/Harman, ready to be interpreted. This is what can enable me to grind through many speaker measurements. What others do is fine for one review every other month but not the style of work and rate of publication we have here (daily reviews).

 

[JohnYang1997]

It's hard to say. It's much harder to do speaker measurements. I would personally say, stick with electronics. And probably let someone else assist doing it probably in different locations or something.
Also the thing is speaker performance is much depending on the room. The proper next step would be headphones to me.
Being said, I'm not discouraging. If that's possible that's really good. And i am looking forward to it.

 

[JohnYang1997]

Oh, we can measure those just as well. Here is Harman's measurements of the ML:

Wow, would they not choose some speakers have equally good frequency response and do it that way? The last three examples are obviously too ****** to sound any good.

 

[Biblob]

How about a GoFundMe or IndieGoGo etc.?
If it's about getting bulk money, I guess this might be most effective in getting it from the visitors.
Or maybe make some deals with speaker manufacturers to let the Klippel be shared or be used for own use. Or would be this to constraining to stay indepent?

 

[napilopez]

It's a weird coincidence I'd just happened to read about this system for the first time earlier today.

My vote is absolutely yes @amirm. I write some audio reviews for a non-audio site and it was ASR that inspired me to change my focus and try to provide objective data within my financial means and current knowledge. It also taught me I should pretty much not bother trying to interpret the sound of amp/dacs/whathaveyou =]. These are things I and most audiofools had long known to some degree, but the data just makes it that much more blatant. I say all this simply to point out that your work has had a meaningful impact for many an audio enthusiast and has likely shifted the industry altogether.

While the Klippel system looks fascinating and it would be amazing to start off with such a strong tool, I do not know if that's something you necessarily need to get the ball rolling. I suppose it depends on what your goals are.

Your electronics measurements are second to none using the best equipment available, so I understand wanting to reach the same standards for speaker measurements. On the other hand, it seems to me the primary goal is to provide more comprehensive data to the public on a wider breadth of speakers than is currently available. The bar for published speaker measurements is not very high.

While yes, there are far more publications measuring speakers than electronics, I, for one, don't think there are nearly enough. For the most part, existing measurements are for expensive hi-fi products. But there are a myriad of other products that deserve to be measured more often. Affordable products, active speakers, small internet-direct brands, and heck, there are even 'lifestyle' speakers that offer surprising results. The Sonos Play One has better spins than some far more expensive speakers. Even 'just' armed with a lazy susan and a decent mic, you could probably cover a wide breadth of speakers normally ignored, and build up the funds for something like the Klippel over time.

But, certainly, it would be amazing (and presumably more time efficient) if the Klippel can be acquired from the get-go.

 

[KSTR]

The Klippel system solves this problem. It makes measurements at two distances from the source and uses that to compute the reflections and subtracts them out.

Yes, but this requires a extreme level of precision matching of the channels and excellent signal-to-noise ratios to work properly, like any other subtractive technique. You still should be away from the DUT at least 5x the largest DUT dimension for a representative wavefront to build up and that alone makes channel signals almost the same unless your channel spacing is also large. And still as much reflection damping as possible is of benefit, though not to full anechoic levels.

 

[Krunok]

Yes, but this requires .. excellent signal-to-noise ratios to work properly

What a lovely reason to tell wife to be quiet!