Speakers that measure well and work in small room

[garbulky]

It’s true that my sited anecdotal observations don’t invalidate blind listening tests or psychoastic principles.

I’m just trying to get my mind around the root of my perceptual paradox. I have a few theories, my main contenders are:

—My subject preference is for speaker designs that date to the time I grew up, and which my taste in music was formed. (Which remainec surprisingly stable, resistant to styles of music popular before and after the 1970s). Music is a meaningful experience, which involves extensive cultural influence and learning. The sound of the reproduction devices themselves are inextricably linked with the meaning of the music. (As well as visual and design elements.) I don’t think dismissing this as nostalgia acknowledges the significance of the experience.

— My musical taste reflects recordings that were created on speaker designs of the era. My thesis is that similar designs better represent this music. They are more “accurate” to the “sonic” signal, which is the medium upon which the artist create, not the electronic signal. (Working on “colored” speakers encodes a kind of “inverse signature” into the signal)

— Most of my listening to the kind of “accurate” speaker designs which objective evidence points to as being generally preferred has been in recording studios which present a very non-typical listening environment. They are treated to heavily control reflected sound, and to absorb low frequencies. Attempts are made to decouple the speakers from the physical structure, which leads to less subjective experience of the bass frequencies.

Conversely most of my experience with “hi-fi” speakers has been in untreated homes. Perhaps if the experience was more evenly mixed, my subjective preference would change.

— That “boxy” sounding speakers provide a final level of integration for poorly integrated multitracked recordings, with the physical nature of the box adding an element of “realism” to the sound. I’ve heard the sentiment expressed that a truly “good” loudspeaker will clearly separate the “good recordings” from the bad. I question whether this is truly an ideal in a world where “bad recordings” outnumber the good.

The Earl Geddes quote reflects a point of view that I think is unsupported, verging on non-sensical. When people talk about audio equipment being “accurate” it generally refers to a narrow perspective on wether a non-transducer device preserves the integrity of an electronic signal.

The reason that the psychoacoustic research was able to apply this more generally to the realm of speakers was because this research showed that listener preferences corresponded with speaker designs that could accurately reproduce electronic signals in a way that allows another transducer to recapture the signal. The basis of the research is the subjective responses, not the “accuracy”, and it is the reproducibility that gives confidence that this is an objective phenomenon. This is the entire foundation that allowed careful measurements to be made that teased out the measurable components of the effect. Without this research on subjective preference, there would be very little meaning in calling a three-dimensional transducer “accurate” compared to a one-dimensional electronic signal.

As someone who is both interested in my personal experience with audio, and the larger, very rich, very complex, multi-subjective, cultural experience of audio reproduction, this view of accuracy has an uncertain role. Ironically, the debate between “objectivists” and “subjectivists” is a red herring when it comes to considering the real experience of listeners. This debate reflects the narrow concerns of neurotic audiophiles, who become convinced that the meaning of their experience is dependent on inaudible differences in audio equipment.

Ironically, considering the results reported by the research of people like Toole and companies like Harman, the real world listening environment is deviating at an ever rapid pace from the traditional notions of “high quality” stereo reproduction.

My step children listen to music generated largely by entirely non-acoustic, non-electronic, signals. Their most common listening environment are Bluetooth earbuds. One of them will also walk around holding a little Bluetooth speaker with some kind of bass enhancer in his hands. Then there is the car, in which they inevitably will crank the bass given a chance. It’s only because I have an actual stereo system setup in front of the TV that they even hear such audio, which is mainly when they play video games. They never notice when I change the speakers!

Add in car audio. Add in speakers on phones, laptops, TVs. True multichannel audio does not appear to be taking off, instead we are getting crazy sounding sound bars which are using DSP to attempt to create 3D sound fields.

On the production side, accuracy in digital audio represents a dilemma. While useful for ensuring storage and copying of data, it has notabil limitations as an artistic medium. This has brought unprecedented development in DSP distortion in attempts to introduce the non linearity of legacy mediums.

My point is that the obsessive focus on accuracy in things like amps and DACs is a sideshow at best, and meaningless in most context because the differences aren’t audible, or if they are it doesn’t matter. My subjective experience is the attempts to apply this type of critical analysis to loudspeakers are frought challenges.

I don’t accept that my subjective preferences in loudspeakers are akin to nutjobs who insist that their “tube rolling” is the magic ingredient that cause their esquisite playback systems to generate a “transcendent listening experience, subtle yet palpable, capable of communicating, the cosmic intent of the artist in a truly musical way.”

If your subjective listening is exactly how you would normally use your gear, then I suggest that perhaps that is it's very stregnth. DBT can tease out a more objective position, show you that something was interfering with your perception. But if your experience with your geara is different to your DBT experience, which one do you think is more helpful? There is no one answer - many will say DBT because it helps them "get over their bias". I say subjective listening. Also I agree with your other point. What we hear isn't magical. But if there is something audible that is causing sound differences, then it must be reflected there. So we need to concentrate on the audible ranges - the ones that appear perfect by measurements. It makes sense to me that this is where it's at. Where's the difference there? Is there another way to look at it we have not considered? I have no clue.

 

[Willem]

Much depends on how large or small you want them to be. I use a pair of Harbeth P3ESRs as desktop speakers in my study and I am more than happy with them as near field speakers. I have also experimented with them in a much larger room, and supported by a subwoofer. Integration was excellent (using an Antimode 8033 dsp room eq unit), but with their small driver they could not quite fill the large room. If you can tolerate the larger size of the Harbeth M30.1 they would certainly play nicely in a room such as yours.

 

[HammerSandwich]

Most of these synthetic sounds are generated by virtual instruments in the computer. So, unless a recording went through an analog mastering (or mixing) process (which many still do as a final step) what you are hearing is data-data-data...data-sound.

I can understand this but not how it's "non-electronic."

 

[andrew]

It looks as if I'll get a chance to listen to the KH310a later this week albeit set-up in a monitoring environment. I'm also intrigued by the Hedd Type 20 due to the ribbon tweeter, great pro-audio reviews and lineariser VST plug-in. That said, I don't have the option to listen before purchase, ham a bit concerned about off-axis measurements (as haven't seen polar response measurements) and I'm not sure that I can use the linearuser@plugin when using Audiolense. Any feedback on the Hedd Type 20?

 

[andreasmaaan]

It looks as if I'll get a chance to listen to the KH310a later this week albeit set-up in a monitoring environment. I'm also intrigued by the Hedd Type 20 due to the ribbon tweeter, great pro-audio reviews and lineariser VST plug-in. That said, I don't have the option to listen before purchase, ham a bit concerned about off-axis measurements (as haven't seen polar response measurements) and I'm not sure that I can use the linearuser@plugin when using Audiolense. Any feedback on the Hedd Type 20?

There are some very basic horizontal polar response measurements here:

 

[MSNWatch]

How about vertical response deviation for the Hedd type 20s?

 

[andreasmaaan]

How about vertical response deviation for the Hedd type 20s?

I haven't seen it measured anywhere but it can be inferred from the dimensions, crossover frequency, and driver topology.

According to that article I linked, the centre-to-centre distance between the midrange and tweeter is 13.4cm and the crossover frequency is 2.5KHz. That should translate into nulls about 30° vertically off-axis, with reasonably even response +/-15° or so off-axis. Looking at the step response, the xover slopes seem to be approximately 4th order, which should translate into a reasonably narrow bandwidth for the null.

The ribbon tweeter is likely to have a more chaotic vertical dispersion pattern in the high frequencies than a dome with equivalent surface area would, but is nevertheless quite small and therefore probably not too chaotic until quite high in frequency.

In other words, overall the vertical polar response is likely to be reasonably good - quite comparable to other well-performing monitors.

 

[andrew]

There are some very basic horizontal polar response measurements here:

View attachment 23814

How would one interpret this figure in terms of real world performance given the difference between the off-axis on HF vs LF side?

 

[andreasmaaan]

How would one interpret this figure in terms of real world performance given the difference between the off-axis on HF vs LF side?

To be honest, it really puzzles me. The woofer-mid cover is at 250Hz. The horizontal pillars should be fairly symmetrical at higher frequencies.

On reflection, I wonder if that graph is actually the vertical polar response and has been mislabelled.

 

[Juhazi]

Vertical off-axis would have a dip around 2.5kHz? Perhaps the speaker's rotation axis was in the middle of the speaker box, not at mid-tweeter axis.

 

[andreasmaaan]

Vertical off-axis would have a dip around 2.5kHz? Perhaps the speaker's rotation axis was in the middle of the speaker box, not at mid-tweeter axis.

Yes, or perhaps the design axis is not quite on the tweeter axis. Hard to say when it’s not even clear whether it’s the horizontal or vertical polar response in the first place..

 

[andrew]

There are some very basic horizontal polar response measurements here:

View attachment 23814

The review from which this was sourced seemed to indicate that this was horizontal making a point of noting the differences between the HF and LF sides of the speaker. It's interesting to note, though, that the KH310a polar response measurements don't seem to have any obvious differences to the LF / HF sides.

 

[Juhazi]

What is obvious then, but L/R offaxis are diffrent for Neumann KH310a - look at response 400-5000Hz. If this is a problem, one must choose a different model.

 

[March Audio]

I
I also think subjective experience is indeed the ultimate judge of an audio system. The foundation of the Toole research are such experiences, it just happens that group preference aligns with notions of "accuracy" in the electrical domain. It seems entirely plausible to me that processing devices, whether speakers or otherwise, could be created that would decrease accuracy and increase subjective listening pleasure.
.

Indeed they are the foundation of the Tool research. However it doesnt "just happen" that the preference aligns with notions of electrical accuracy. Its not an accident. It is clearly demonstrated that preferences do align with acoustic accuracy.

With respect, your sample of one personal ideas don't really change this fact.

 

[Ron Texas]

[QUOTE="Sancus, post: 160409, member: 4609" All the wealth is owned by older people.[/QUOTE]

You have to be careful when making generalizations in a public forum. There are many people who became rich young from working for Microsoft and other big tech companies. Older people tend to have wealth accumulated because they had more time to earn, save, and due to the compounding of returns. Remember, on the average senior citizens are struggling financially because they have past their peak earnings years, have high health care expenses and were unable (or unwilling) to save for retirement.

I often make this comment, think about how you are going to retire before spending large sums on your hobbies, whatever they are. Europeans often answer they have a better version of Social Security than we have in the US.

 

[andrew]

What is obvious then, but L/R offaxis are diffrent for Neumann KH310a - look at response 400-5000Hz. If this is a problem, one must choose a different model.

You're right there is some left / right anomalies. I find it difficult to get the magnitude in this form of plot. When, though, would it not be a problem? I guess if the room is very large or heavily damped?

 

[andreasmaaan]

The review from which this was sourced seemed to indicate that this was horizontal making a point of noting the differences between the HF and LF sides of the speaker. It's interesting to note, though, that the KH310a polar response measurements don't seem to have any obvious differences to the LF / HF sides.

Yes they say that, but I strongly suspect their idea of the speaker’s normal orientation was confused.

 

[b1daly]

What do you mean by 3D measurements?

I tend to agree with this, but would have thought it was the job of the engineers producing the recording to include such distortions. Perhaps you’re simply saying that your taste is for a higher level of distortion than that preferred by most engineers? There’s actually quite a bit of experimental evidence supporting the view that most listeners prefer additional distortion than that provided by a typical recording / accurate playback system.

IMO these “systems” are in fact the effects units and other processors that engineers use in the recording/mixing/mastering process.
Why couldn’t a controlled study be based on long-term listening? But in any case, there is overwhelming evidence suggesting that if something is not audible with very short-term listening, it is not audible with longer term listening.

I wasn’t sure how to intersperse my answers to your well put questions/comments on my phone where I’m typing this, so I’ll just reply in order.

By 3D measurements I simply mean measurements that address multiple angles and distance from the speaker, like the Harman spinorama tests.

I agree that it is the job of of the producers/engineers to introduce beneficial distortion that helps integrate the recordings. There are a couple of limits with this.

One is that it is hard to do well, especially in the digital domain. So the speaker is the “last chance” to pull a poorly integrated mix together. This is just a personal theory, but in the debates in the production world about the advantages/disadvantages of digital audio, I think the place where it is most impactful is in the mixing. In an analog mixing desk, there are complex, non-linear, basically chaotic interactions between the signals being mixed, due to the various suboptimal performance of the mixing device. In particular crosstalk and the inherent non-linearity of analog amplifiers. As more signals are added into a bus, pushing the amp into a different gain range, the sound of one instrument affects another.

There have been attempts to capture these effects in DSP algorithms, but so far I think it is work in progress. One issue is that most dsp approaches are deterministic, where the same input gives the same output. In contrast analog devices are non-deterministic, especially something complex like a mixer, and can’t even produce the same output to the same input.

So my theory is that a speaker with a somewhat resonant box provides a layer of non-deterministic signal processing, which my ear recognizes as “real.” This I think is a subjective preference, and haven’t heard this shared by many.

Interestingly; my studio partners subjective experience largely matches my own. Of the multiple monitors we have in our studio, we share the perception that only one pair will suffice for “fun listening.” This is a Quested 108b, which is an old design, rectangular box of mdf, 8inch woofer w dome tweeter, passive crossover.

In extension, everyday listening rooms, undamped on the physical structure or with reflections do provide a similar chaotic playback environment. Often wildly so, with a huge variety of listening and speaker placement issues within a single room, and across rooms.

The problem with this, or any kind of distortion introducing element in a playback system, is that it is hit or miss whether it will flatter or not the particular recording.

A question I have for, and this gets to the heart of my theory about my own perceived subjective preference.

Most of the classic rock recordings from the 60s to 2000s were mixed on wooden boxes. That is as close to the “true signal” as anything. (The reality is most producers/engineers work hard to make mixes that translate across a wide range of playback systems, which is why the notion of “accuracy” is hard to apply outside of the electrical characteristics of things like DACs and amps).

So if we want to bring a good representation of this experience into our home, my thesis is that a wooden box has a better chance of achieving this than a theoretical perfect speaker that would deliver only an accurate representation of the electronic signal.

The problem is is that the playback experience of the producers, especially of idiosyncratic studio monitors is not encoded in the signal.

To capture and deliver such an experience would require a system that could create the illusion that wooden boxes were playing in your room! I don’t think this is possible, especially when it comes to how we localize sound as we move in space.

Even the a mixer and mastering engineer will try to transcend the specific monitor system the album was mixed on, this is not entirely possible to do.

Because on a multitrack album, as the artists work, they respond to what they hear on playback, creatively. This means the sonic signature of the monitors can never be fully “backed out” of the signal.

To anyone wondering what the hell I’m talking about think about classic rock albums like Back In Black, Rumours, Dark Side of the Moon. My contention is that playback on a boxy speaker will be more faithful.

 

[andreasmaaan]

Most of the classic rock recordings from the 60s to 2000s were mixed on wooden boxes. That is as close to the “true signal” as anything. (The reality is most producers/engineers work hard to make mixes that translate across a wide range of playback systems, which is why the notion of “accuracy” is hard to apply outside of the electrical characteristics of things like DACs and amps).

So if we want to bring a good representation of this experience into our home, my thesis is that a wooden box has a better chance of achieving this than a theoretical perfect speaker that would deliver only an accurate representation of the electronic signal.

Thanks for the interesting perspective, and I think there's logic to what you're saying, but I disagree with this point.

What I find impossible to come at is the idea that there is a specific (or a specific range of) "boxy speaker" sound, and that this is what is missing from accurate monitors.

Boxes can do all sorts of arbitrary things to the output of a speaker. In particular:

• Baffle dimensions and box depth determine the box's effect on the speaker's polar response. If you want to recreate a specific "boxy speaker" polar response to be consistent with how the music was supposedly intended to sound, wouldn't you need the box to have specific dimensions similar to those on which it was originally mastered?
• Box resonances similarly vary from box to box - indeed far more so than baffle dimensions. One box will create an entirely different set of completely arbitrary box resonances than another. How can replacing one arbitrary set of resonances with another entirely arbitrary set of resonances be more faithful to the recording?

In the end, I think that if you prefer the specific colourations of a box speaker, that's because you like that speaker's colourations - not because it gives you a more faithful representation of the original recording.

There have been attempts to capture these effects in DSP algorithms, but so far I think it is work in progress. One issue is that most dsp approaches are deterministic, where the same input gives the same output. In contrast analog devices are non-deterministic, especially something complex like a mixer, and can’t even produce the same output to the same input.

So my theory is that a speaker with a somewhat resonant box provides a layer of non-deterministic signal processing, which my ear recognizes as “real.” This I think is a subjective preference, and haven’t heard this shared by many.

This is an interesting possibility I think, and the aspect of your discussion that I find a lot more convincing.

I don't know enough about analogue electronics, but I would have thought "non-deterministic" might be going a bit too far? What do you mean a mixer can't reproduce the same output to the same input? That if you turn it on one day and play the same recording through it, the output will be significantly/audibly different than it would be on the next day?

In terms of speakers, I would actually say that boxes tend to be among the more deterministic of speaker components, since box resonances tend to be determined by materials, structure, and dimensions, etc, all of which remain relatively stable regardless of external factors. Box damping materials may behave less deterministically as they tend to heat up to a more significant extent than MDF, I would say. Drivers and passive crossover components are actually likely to behave far less deterministically than boxes IMHO. Perhaps this behaviour in passive crossovers is something that tends to audibly distinguish them from active speakers..

 

[Ron Texas]

Yes they say that, but I strongly suspect their idea of the speaker’s normal orientation was confused.

I spend a lot of my time confused regardless of orientation. It is safer to be horizontal as confusion in the vertical orientation may result in serious injury.

By the way, has anyone ever knocked on the "box" of an LS50? That baby is solid, no matter what you might think of the speaker.