Is there any reason to still be looking out for strict linearity nowadays?

[anphex]

I believe with the current consumer technology you can spend a little money for measurement equipment to get about any speaker to a ruler flat response on your desired listening position.

After thinking a while, the following ones are the crucial characteristics to look out for in my opinion:
- lowest possible distortion, most importantly in low end - usually made possible with large, heavy enclosures and lots of piston area
- may correlate with the upper characteristic but also: far reaching bass extension
- wide and smooth sweet spot
- short and smooth decay of the whole FR

As long as these fixed values are fine, you can EQ your system to try to achieve 1dB linearity, sacrifice a little max SPL and distortion and have a reference system.
So, as long as you have a miniDSP, AVR or even just a computer the former is pretty easy as long as you have an USB-measurement mic or a spare interface with a xlr measurement mic.

Of course this all only applies to people who care about linearity and want to go through the hassle. So factory-linear speakers will always have a customer base.

Edit: I remember that with the newer miniDSP and some amp you can "activate" any speaker with up to three ways and optimize its performance even beyond the provided passive crossover. You unplug the passive crossover, wire the chassis directly to the amps. The amps to the miniDSP and let the software do it's thing. Boom, hackjob active speaker. Though hackjob may sound mean considering the probably far better result compared to passive.

 

[fpitas]

It really depends on the speaker if simply EQing to be ruler flat at the listening position is a viable action. With some speakers you'll just create a sonic disaster, or at least make things worse. If the dispersion characteristics are good, then you usually can.

 

[Sokel]

Who can stand ruler flat at MLP?

 

[fpitas]

Who can stand ruler flat at MLP?

I *assume* we're talking time-gated. But good point.

 

[fpitas]

I believe with the current consumer technology you can spend a little money for measurement equipment to get about any speaker to a ruler flat response on your desired listening position.

After thinking a while, the following ones are the crucial characteristics to look out for in my opinion:
- lowest possible distortion, most importantly in low end - usually made possible with large, heavy enclosures and lots of piston area
- may correlate with the upper characteristic but also: far reaching bass extension
- wide and smooth sweet spot
- short and smooth decay of the whole FR

I agree with your wish list. However, the trick is achieving all of that. You'll need room treatment too, almost certainly.

 

[anphex]

Of course a good room is needed, but let's imagine an ideal room for this one.
Also, yes, ruler flat sucks, but when you use it as a basis for a automatic louness curve it's probably as good as it can get.
Apart from that you should probably use a fixed loudness curve - which is also linearity with humps in the low and high end.

 

[fpitas]

Many people would give a lot to have an ideal listening room Then it's down to the dispersion characteristics, the distortion at reasonable listening levels, etc. Perhaps the power compression too.

 

[HarmonicTHD]

Yes. Anechoically flat (not flat at MLP) is still very much desirable plus ideally “good” (constant) directivity and low distortion at max SPL (Harman studies).

If not, what are the scientifically sound alternatives?

Will that make all speakers sound the same? No, because one can still have differences on the dispersion angle plus all the differences due to the various rooms and different max SPL capabilities.

 

[sergeauckland]

Edit: I remember that with the newer miniDSP and some amp you can "activate" any speaker with up to three ways and optimize its performance even beyond the provided passive crossover. You unplug the passive crossover, wire the chassis directly to the amps. The amps to the miniDSP and let the software do it's thing. Boom, hackjob active speaker. Though hackjob may sound mean considering the probably far better result compared to passive.

Indeed, which is what I've done with my 801s. I think that any good passive 'speaker can be improved by going active and equalising accurately using a DSP-based crossover and measuring software. If done sympathetically, i.e. removing the crossover and keeping it safe somewhere, it's perfectly possible to reverse the process should be 'speakers be sold. Or put the crossover into a box, and one can play active vs passive games.

Software like REW has made setting up a two or three (or more) way crossover and equaliser easy if tedious, the only real limitation for most people is equalising the bass as gated measurements only work down to around 200Hz due to room dimensions. However, bass is very affected by room conditions, so in-room EQ for bass would be of more use than anechoically flat, so the gated measurements LF limitation is less of a concern.

There are a (very) few manufacturers who supply external crossovers, and these tend to be the ones that one wouldn't want to use if a flat response is preferred, but it's an opportunity I think for manufacturers more into the installed Home Theatre market to exploit.

S.

 

[pma]

Yes. Anechoically flat (not flat at MLP) is still very much desirable plus ideally “good” (constant) directivity and low distortion at max SPL (Harman studies).

If not, what are the scientifically sound alternatives?

Will that make all speakers sound the same? No, because one can still have differences on the dispersion angle plus all the differences due to the various rooms and different max SPL capabilities.

You are right. The idea of having poor speaker drivers and poor box material all corrected by DSP is incorrect, at least in case that one seeks for high quality sound reproduction.

 

[Daverz]

Who can stand ruler flat at MLP?

One fits to a target curve with a falling treble. I use a target curve that is flat to about 200 Hz and then falls 6 dB by 24kHz.

 

[Sokel]

One fits to a target curve with a falling treble. I use a target curve that is flat to about 200 Hz and then falls 6 dB by 24kHz.

For me is +3db at 30Hz descending to at least -3db at 20Khz centered at 200Hz.
For low late night listening lows are even higher.And I can't stand even the slightest raise at around 2-5Khz,I'd rather listen to a nice chainsaw cutting through steel.
But that's only taste.

 

[HarmonicTHD]

One fits to a target curve with a falling treble. I use a target curve that is flat to about 200 Hz and then falls 6 dB by 24kHz.

This is a so called preference curve. Such as the Harman Preference curve, which derives it selves from a flat anechoic speaker frequency response.

And because preferences vary individually within a margin of the Harman curve one should adjust the in-room target curve to one’s individual preferences. That does not however negate the desirable target of a flat anechoic response for speakers as a foundation.