Harbeth 30.2 vs kef R3

[Koeitje]

i m ass well addicted to fun and learning
concerning electronics dac and amp what is the point can we say sinad chart for dac 100 for amp 80
and for speakers ?

https://www.audiosciencereview.com/...-thresholds-of-amp-and-dac-measurements.5734/

 

[Pjetrof]

Thx will have a read

 

[Absolute]

In my experience speakers that measures better also sounds better, which is not surprising given the research. The main factor for sound quality in speakers relates to frequency response both on and off-axis.
My experience is that speakers with equal frequency response will sound extremely similar given similar dispersion, so I've yet to discover a magic sauce that renders frequency response (including dispersion) insufficient to explain most of the perceived differences.

For this reason it seems very sensible to find speakers with the best frequency response and directivity that suit your needs when it comes to size, spl capability, looks etc and then just make it sound however you want/like by tone control/EQ.

If no EQ is allowed, there's a strong possibility that the Harbeth will sound more relaxing/better over time than the R3 which has an unnecessary broad-band rise in the higher frequencies. This is one of those things that doesn't come down to quality per se, but relates to tuning and hence individual preference.

Personally I have more faith in IMD distortion than THD as a factor when frequency response is adequately similar when it comes to perceivable differences, but can't say for sure.

 

[Willem]

I have two observations:
1 The Harbeth M30 that was measured here is an older model, a few generations ago, so the measurements do not necessarily reflect the current incarnation.
2 All speakers are inevitably very imperfect, so the analogy with small or non-existent differences between electronics is quite pointless. Of course, every speaker designer tries to push the envelope, but they all have to accept compromises and one designer's best compromise may not be another's, let alone to your liking. My personal preference is for Quad electrostatics, but they have real limitations. In small dynamic speakers, my personal preference is the Harbeth P3ESR, but like every other speaker, it is not perfect.

 

[Pjetrof]

Thx everyone for the responses, I think I asked my question wrong with speaking about the harbeth and the kef.
the answer from @Absolute is what I was looking for. Like electronics even the diffrences are so small they are inaudible but the best measured one is the best.
Does this apply as well for speakers, just like an amp have to reinforce a signal without adding or distracting something,
is it fair to say a speaker may as well not ad or distract something of the signal.
there fore the best measured speaker is the best speaker?

 

[thewas]

Does this apply as well for speakers, just like an amp have to reinforce a signal without adding or distracting something,
is it fair to say a speaker may as well not ad or distract something of the signal.
there fore the best measured speaker is the best speaker?

The problem is that audio electronics cope only with one dimension (that is amplitude as a function of time) which in the last decades is easy to make inaudibly perfectly linear and with inaudible distortions but loudspeakers output in a 3-dimensional space and even if you would manage to make their output very linear and low distorted in one direction (which is already very hard) you never have the same exact output in other directions, as all different driver, baffle and crossover configurations give different radiation patterns.

Also there is no total agreement on the perfect radiation pattern, only general concepts depending on room acoustics, listening distance, recording technique and listening configuration (mono vs. stereo vs. multichannel). So, even if you theoretically could actually listen in an anechoic space to one or more loudspeaker that don't "add or distract something to the signal", even that wouldn't be the desired solution as it would rather sound like a headphone, as almost all available recordings are made and optimised to be listened in a normal room with reflections where you also hear not only the direct on-axis sound of the loudspeaker but also its radiation to all other angles through room reflections.

 

[Archsam]

As some of the above replies have already mentioned, unlike other electronics such as amps and DACs, a speaker has to interact with the physical room it is placed in, and how the drivers / enclosure interact with each room is unique, and the sound you perceived will depend on the speaker set up and whether room correction is applied or not.

A speaker that measures well will have a much better chance in sounding good, but that is likely only half the story.

There is also the scale factor of the room vs. speaker - for example, in my smallish living room the distance between my speakers is 1.7m, and the distance between my ear and the tweeter is 2.7m. My tiny Harbeths sits 35cm in front of the wall, and sound really good in this space / arrangement. 80% of the time I don't even turn on my subwoofer (though others might prefer more bass). I am pretty certain that if you put a pair of Revel Salon 2 in their spot I will not be getting the sound they are capable of, or even an enjoyable set up due to the lack of space around the speakers / insufficient space to let them create a good soundstage / overwhelming bass due to the multiple drivers overpowering my room.

 

[Ron Texas]

Not necessarily. The score is based on average listener preference. This means that more listeners will like a higher scoring speaker. It does not mean that every listener will like a higher scoring speaker.

The preference score is a perfect tool for commercial manufacturers who want to sell as many speakers as possible. It is not perfect (though helpful) for the individual buyer.

Thank you for a distinction without a difference. It's far fetched to interpret any score based on a group of people as being unanimous.

It's debatable whether the preference score is a perfect tool for manufacturers. Klipsch and B&W enjoy widespread success with speakers which are intentionally voiced bright because on a quick listen they stand out and sell. The result is disingenuous, but so are a lot of things.

 

[avanti1960]

In my experience speakers that measures better also sounds better, which is not surprising given the research. The main factor for sound quality in speakers relates to frequency response both on and off-axis.
My experience is that speakers with equal frequency response will sound extremely similar given similar dispersion, so I've yet to discover a magic sauce that renders frequency response (including dispersion) insufficient to explain most of the perceived differences.
.

One key "secret sauce" that renders frequency response as woefully inadequate in determining the sound is driver materials.
With identical frequency responses different speaker system using woofers (for example) made of paper, aluminum, ceramic, carbon fiber, poly, bamboo etc. will all have significantly different sound characteristics. Same thing for midrange drivers, tweeters and even subwoofers.
Different compounds, coatings and material thicknesses will also affect pitch, timbre and tonal accuracy in addition to driver base material.

 

[MattHooper]

One key "secret sauce" that renders frequency response as woefully inadequate in determining the sound is driver materials.
With identical frequency responses different speaker system using woofers (for example) made of paper, aluminum, ceramic, carbon fiber, poly, bamboo etc. will all have significantly different sound characteristics. Same thing for midrange drivers, tweeters and even subwoofers.
Different compounds, coatings and material thicknesses will also affect pitch, timbre and tonal accuracy in addition to driver base material.

I'm not sure the science actually supports those claims.

Admittedly I myself have felt a pull, like many audiophiles, towards presuming certain drivers or materials have a "sound" in speakers, but if I understand (just barely!) what I've read, so long as a driver's resonance is well suppressed, it shouldn't be a factor in coloring sound. (I seem to remember Floyd Toole commenting on this at one point).

Anecdotally, I've had my intuitions about driver materials overturned numerous times.

For instance, my Thiel 2.7 speakers (and previous 3.7s) are all metal drivers (aluminum diaphragm midrange, and tweeter, woofer metal as well). They do not sound remotely "metallic." In fact, the reason I ended up trying and selling my Harbeth SuperHL5plus speakers is because in direct comparisons, the Thiels sounded so similar in terms of timbre/tonality (that "organic" sound), but did so with greater precision and realism.

I used to have an intuition that metal tweeters did metal sounds better - drum cymbals etc sounding more properly metallic. But the soft dome tweeter of my Joseph speakers excel in just that area.

I can't say it's easy to totally shake the intuition about driver materials producing distinct timbral character, but I've yet to see the idea strongly backed by science and it's not perfectly consistent with my own long experience with speakers.

 

[thewas]

With identical frequency responses and distortions lower than the perception limits drivers from different materials actually sound the same but only if you "switch off" their different radiation pattern, i.e. listening to them in free field conditions, see this research work https://www2.ak.tu-berlin.de/~akgroup/ak_pub/abschlussarbeiten/2010/RotterAndreas_MagA.pdf from which I electronically translated to English the conclusions:

The present paper investigated the widely held opinion that the subjective sound impressions of tweeters are clearly related to the drive principle, the diaphragm material used and the geometric extension of the diaphragm. Six tweeters, typically used in current loudspeakers, were selected for this purpose. They differed both in the transducer principle, their geometry and in the sound-generating material of the diaphragm.

In order to exclude typical discolouration of the sound image due to interaction with the surrounding room as well as with the baffle of a loudspeaker cabinet, all drivers were examined in a quasi infinite baffle in a reflection-free environment. In order to avoid discoloration of the sound image resulting from driver-specific distortions of the amplitude and phase response alone, all drivers were equalized for one point in the room using a digital crossover based on FIR filters. The aim of the compensation was a linear amplitude response in the range of 2 kHz to 22 kHz with maximum deviations of ±1 dB as well as a linear phase response.

After successful equalization of all drivers, binaural impulse responses were recorded by them in combination with an equally compensated bass-midrange driver. The BIR was measured in positions relative to the drivers that corresponded to a left and a right speaker in a typical stereo arrangement. To allow a dynamic simulation of the loudspeaker pairs in a later comparison, the BIR were plotted in a radius of typical head movements.

In a final listening test, the two virtual loudspeaker pairs that showed the greatest deviations from each other when analyzing the binaural data were compared. Using the binaural, dynamic simulations of the loudspeaker pairs, ten test persons were asked to find out whether there was still a difference between the tweeters that could be determined by the subjective sound image. The result of the experiment allows the conclusion that there is no perceptual difference between the tweeters. In conclusion, it can be stated that in the case of excluded interaction with the room and the baffle, with axially linearized drivers in amplitude and phase, there is no longer any difference in linear operation between loudspeakers with different transducer principles, different membrane material or expansion. The common saying can thus be considered disproved, at least under these conditions.

Further binaural measurements and simulations can now be performed to determine the effects of other parameters such as the directional characteristic and the resulting driver-specific interaction with the room. Likewise, the differences resulting from driver-specific interactions with the baffle or from the operation of the tweeters in the non-linear range can be investigated.

This work thus busts the common audio myth of different sound of membrane materials that is not described by the usual metrics like frequency response, radiation pattern and distortions, although on the other hand in real life two or more drivers with same FR and identical radiation pattern don't really exist.

 

[Ron Texas]

One key "secret sauce" that renders frequency response as woefully inadequate in determining the sound is driver materials.
With identical frequency responses different speaker system using woofers (for example) made of paper, aluminum, ceramic, carbon fiber, poly, bamboo etc. will all have significantly different sound characteristics. Same thing for midrange drivers, tweeters and even subwoofers.
Different compounds, coatings and material thicknesses will also affect pitch, timbre and tonal accuracy in addition to driver base material.

Unobtanium produces the most desirable sound.

 

[CDMC]

It think that one thing that may be confusing the OP is his lumping Amps, DACs, and Speakers, as electronic devices. Amplifiers and dacs are electronic devices. Speakers are an electromechanical device, converting electrical energy into mechanical energy.

With electrical devices, we have very quantifiable measurements that are highly correlated and inform us of their qualities. With speakers, the audibility of the different measurements we make become less correlated, in large part because every room is different, placements within rooms different, resulting in vastly different sound, even for the same speaker, when moved from room to room. We know that in general, most people prefer speakers with a flat frequency response that continues off axis. That is correct in most cases, but say you have a large room that is extremely over dampened and listen from 5 meters away. A flat speaker with smooth off axis response will lack high end response, while a tilted up speaker will sound correct.

So while we can predict with strong certainty if an amplifier or dac will work well (is the SN ratio sufficient, does the amp have enough power to not clip), we cannot predict what room a speaker will be put in, the setup of the speakers, or the furnishings in the room. This is why in home speaker auditions are so important for speakers, but not for dacs, amps or other source components.

Finally, the OP asked if two speakers of the driver design and layout will sound the same. The short answer is sometimes, maybe. You can build two identical speakers, but use two different crossovers, say a 6db/octave versus a 24db/octave, which will give the same on axis response, but entirely different off axis responses, resulting in very different sounding speakers.

 

[Willem]

Unlike electronics, speakers have mass, and hence do all sorts of very complex things such as resonate, or start and stop slowly. So perfection is impossible, and even understanding their behaviour is also much more difficult.

 

[Archsam]

This actually brings up another question I have meant to ask for a long time: a lot of people describes the 'speed' of the drivers, usually by people whose musical preference is mainly rock / metal / electronica.

Harbeth speakers are often criticised for been too 'gentle' and 'civilised', pipes and slippers etc etc, i.e. they are not 'fast. Personally I haven't had an experience where I play a song and think, hmmm this sounds slow.

How do you quantify this characteristic in measurements? How do you measure the start / stop time of a driver, if that's even possible?

 

[Willem]

I think impulse signals can be used to test this. Just look at how well a square wave is reproduced by an electrostat. However, I don't think that when people say a speaker (and even more a subwoofer), is slow their observation has anything to do with the speaker itself. My hunch is that in many cases the real cause of what they perceive is in the slow decay time of room modes.

 

[CDMC]

This actually brings up another question I have meant to ask for a long time: a lot of people describes the 'speed' of the drivers, usually by people whose musical preference is mainly rock / metal / electronica.

Harbeth speakers are often criticised for been too 'gentle' and 'civilised', pipes and slippers etc etc, i.e. they are not 'fast. Personally I haven't had an experience where I play a song and think, hmmm this sounds slow.

How do you quantify this characteristic in measurements? How do you measure the start / stop time of a driver, if that's even possible?

The idea of a driver being "fast" or "slow" is a misnomer. When a driver is asked to produce a frequency, say 100hz, it is asked to move in and out 100 times in a second. If it moves in and out 90 times a second, it is producing a frequency of 90 hz. It it simply doesn't occur. There are two things that do occur (and probably more that I am not aware of) that effect the perceived "speed" of a driver which are almost exclusively in the bass range:

1) Group Delay- This is a lag between the signal and output. Generally it increases as frequency decreases, but the audibility also decreases as frequency goes down. Ported enclosures have greater group delay than sealed enclosures, but any properly designed enclosure will have group delay below the level of audibility. It is generally only very low end speakers and subwoofers that suffer from audible group delay.

2) Frequency Response Variations- These need to be broken into two.

a) The first are response variations of the speaker. When looking at the bass in a speaker, or subwoofer, there is a measurement called QTS which measures the resonance of the system. It is a measure of the linearity of the bass system, with a QTS of .0707 the flattest, higher QTS resulting in a peak in the bass with a faster rolloff, and lower QTS a slower rolloff in higher frequencies (technically this applies to sealed enclosures but vented enclosures exhibit the same characteristics depending on their tuning). A higher QTS results in a "slower" sounding bass system due to the hump in the frequency response, a lower QTS "fast" and or "tight" bass due to no hump, or even a rolloff at a higher frequency but more extended low frequency bass.​

​

b) The second and far more dominate is the room response. When you place a speaker in a room, depending on placement of the speaker(s) and the listener, you can easily see response variations in the under 200hz range of more than 20db. Careful placement can reduce this significantly, which is what you see when people say things like: "I moved the speakers and the bass tightened up." Simply taking a speaker that sounds slow and plodding and measuring it, then using digital equalization to minimize the large variations in the bass, can transform it into a speaker with tight and extended bass.​

​

You will notice again there is a common theme, the room. One thing you will find with high quality professional audio installers, especially on the home theater side, is they are far less concerned about the brand of speakers and differences between them (once they establish sufficient output and a reasonably flat frequency response) than the room itself, setup, treatment (for decay times, using a combination of absorption and diffusion as needed for the mids and highs), listening position and then equalization of the bass frequencies. These room variations are far greater than the speakers themselves and dominate the sound once you get to the level of competently designed speakers.

 

[MattHooper]

This actually brings up another question I have meant to ask for a long time: a lot of people describes the 'speed' of the drivers, usually by people whose musical preference is mainly rock / metal / electronica.

Harbeth speakers are often criticised for been too 'gentle' and 'civilised', pipes and slippers etc etc, i.e. they are not 'fast. Personally I haven't had an experience where I play a song and think, hmmm this sounds slow.

How do you quantify this characteristic in measurements? How do you measure the start / stop time of a driver, if that's even possible?

I didn't find my Harbeths sounded too "slow," but despite that audiophile descriptions are often disparaged, I do understand generally what someone is getting at by describing sound as "fast" or "slow" sounding.

I had a speaker to audition at my place once that I compared to my older Thiel speakers. The new speaker had a thicker, flubbier bass quality and it had a fascinating effect on the perception of the music. I'd play my various demo tracks which include lots of R&B, electronica, funk and jazzy stand up bass-riff stuff, and the new speaker actually seemed to slow the pace of the music! It was bizzare. I'd play a song that starts out with a tight bass riff rythm and then the drums kick in. On the looser, flubbier speaker it literally sounded like the bass player had relaxed and slowed down the pace of his playing. Back on the Thiels it just tightened up and it had a heart-quickening energy and pace. It was like the difference between playing a bass with new strings, vs with old strings detuned for a looser action. There was a sense of the playing/instruments tightening up, becoming taut and faster.

 

[tuga]

With identical frequency responses and distortions lower than the perception limits drivers from different materials actually sound the same but only if you "switch off" their different radiation pattern, i.e. listening to them in free field conditions

Years ago Gradient hosted a comparison between a few speakers playing in an anechoic chamber and the live feed:

More info here -> https://www.inner-magazines.com/audiophilia/an-icon/

 

[q3cpma]

All speakers are inevitably very imperfect, so the analogy with small or non-existent differences between electronics is quite pointless. Of course, every speaker designer tries to push the envelope, but they all have to accept compromises and one designer's best compromise may not be another's, let alone to your liking.

The idea that all compromises are equal because they're compromises is sophism and nothing else.