A collection of speaker target responses in .csv/.txt format

[thewas]

There is some question in the importance of the shape and location of those peaks and dips in these generalized target curves. For instance, there is a tiny hump in the low-mid area of the original extracted curve... does that mean the trained listeners actually preferred just a little bit more oomph in that band? And the dip at 8Khz was that intentional?

Exactly, we shouldn't forget that those curves were made using specific loudspeakers in a specific room, in a specific listening distance and using fixed frequency shelving filters so cannot and should not be generalised as also Toole writes in the first link in my signature.

 

[dasdoing]

One would only need two lines in the text file to create that extremely simplified curve.

There is some question in the importance of the shape and location of those peaks and dips in these generalized target curves. For instance, there is a tiny hump in the low-mid area of the original extracted curve... does that mean the trained listeners actually preferred just a little bit more oomph in that band? And the dip at 8Khz was that intentional?

the whole target thing is sketchy lol
could they reproduce it on another day?
would it be the same if one guy had the flu?
how would it be in another room?
etc..

 

[dasdoing]

in my room the Dr. Floyd Tool target on var smootheing is the most neutral. it results in a more or less flat envelope on the unsmoothed

 

[ernestcarl]

the whole target thing is sketchy lol
could they reproduce it on another day?
would it be the same if one guy had the flu?
how would it be in another room?
etc..

You'd have to review the research design and how they applied their controls.

I don't think it's prescriptive, but rather descriptive -- the trained listeners just happend to like this one more. There is a margin of error so one can certainly deviate here and there -- one could be happy alternating between several presets if available. But it is also the case that one's preferred "house curve" will depend on several factors: some people like more bass, some people want less. Some speakers have resonances in the HF and mids that we may want pull down a bit, and there are systems have severe ringing and/or distortion in the bass even when equalizing flat -- so wouldn't really want to apply too much boost in that scenario. Then there is variability of the source music or soundtracks... when I have a toothache or really bad migraine, or am stressed I really cannot stand HF and find myself shelving it way down.

 

[dasdoing]

You'd have to review the research design and how they applied their controls.

I don't think it's prescriptive, but rather descriptive -- the trained listeners just happend to like this one more. There is a margin of error so one can certainly deviate here and there -- one could be happy alternating between several presets if available. But it is also the case that one's preferred "house curve" will depend on several factors: some people like more bass, some people want less. Some speakers have resonances in the HF and mids that we may want pull down a bit, and there are systems have severe ringing and/or distortion in the bass even when equalizing flat -- so wouldn't really want to apply too much boost in that scenario. Then there is variability of the source music or soundtracks... when I have a toothache or really bad migraine, or am stressed I really cannot stand HF and find myself shelving it way down.

from what I observed those smaller band variations probably depend a lot on how much reflections there are in the room. deep shallow dips (cancelations) result in lowering on smoothed graph as they take the average down. so if there is a range in one room that is particularly reflective, the smoothed graph should have a slight dip there. if you then create a listening preference graph in this room, and then transfer it to another room which doesn't have this high reflective range at the same place, it wont translate. this is how I see it. so a straight (harmann) line should work better as a generalization for the whole audio world.
I think the unsmoothed graph show so much more. at least observe it to see where you room is most reflective (those big shallow dips). there your smoothed graph should have an angle. 0-200 ajust by ear. if the highs are not very reflective, go horizontal there again. then you ajust the amount of angle by ear.
personaly I am using a flat enevolpe target on the unsmootheed graph though.

 

[Presently42]

It seems to me, that target curves have two parts: bass and treble. Bass is room and speaker dependant, and informed by listener preference; whilst treble is mostly speaker dependant (and probably a bit informed by listener preference too). I've had the most success when equalising the bass below the transition frequency range; and finding the tilt of my speakers in my room and equalising them to be smooth to whatever this tilt is. In the end, the result is similar to many curves - but this, I suspect, is mostly a function of the speakers being fairly good.

In other words, as most research (both professional and anecdotal) seems to indicate, not one curve fits all purposes; and looking for such a curve is counterproductive. A curve sounding good in a room is mere coincidence: rather, the curve must fit the room.

But these are merely my musings....

 

[ernestcarl]

a straight (harmann) line should work better as a generalization for the whole audio world.
I think the unsmoothed graph show so much more. at least observe it to see where you room is most reflective (those big shallow dips). there your smoothed graph should have an angle. 0-200 ajust by ear. if the highs are not very reflective, go horizontal there again. then you ajust the amount of angle by ear.
personaly I am using a flat enevolpe target on the unsmootheed graph though.

What exactly is a "straight (harman) line"?

Here I attached an unequalized (*LP and HP filtered only) left and right in-room response measurement. It's fully bass-managed with a subwoofer, level-matched, and time-aligned. How would your straight Harman line and envelope EQ method fit and work in this one example? I know you mentioned that 0-200 Hz can be equalized by ear without reliance on a predefined bass target curve, but, please do indulge my curiosity and make up whatever you think is best... Now it should not be too hard as there are no major dips to worry about and reflections are controlled.

*Oops! wrong calibration file used previously... switched it to 90 degrees

 

[dasdoing]

What exactly is a "straight (harman) line"?

Here I attached an unequalized (*LP and HP filtered only) left and right in-room response measurement. It's fully bass-managed with a subwoofer, level-matched, and time-aligned. How would your straight Harman line and envelope EQ method fit and work in this one example? I know you mentioned that 0-200 Hz can be equalized by ear without reliance on a predefined bass target curve, but, please do indulge my curiosity and make up whatever you think is best... Now it should not be too hard as there are no major dips to worry about and reflections are controlled.

*Oops! wrong calibration file used previously... switched it to 90 degrees
View attachment 169533

what I said is realy confusing, and aplying it to your example is, too (I found)....so forget it.
but what I tried is to create a target theory. let me explain.
this is an anechoic room that suddenly becomes difusive (simultaed with Rephase, there are obviously no modes in this fiction room, just cancelations):

what happens in var smoothing is that we see our slope when we go from anechoic to difusive. but while our disfusion is more or less constant it becomes flat again:

from what I observed the changing "difusivness" of rooms over the FR explains why targets for smoothed graphs are necessary.
If there is no difusion (anechoic room), the var target would be flat (obviously)
but I think in a totaly constant difusive room the target would be flat also

But again, aplying this to create a target doesn't seam to be practicle.
but I will create an equalization with my flat envelope method with your data so we can see what the resulting curve in var is....and maybe you try it out. just sending this post seperatly

 

[ernestcarl]

from what I observed the changing "difusivness" of rooms over the FR explains why targets for smoothed graphs are necessary.
If there is no difusion (anechoic room), the var target would be flat (obviously)
but I think in a totaly constant difusive room the target would be flat also

But again, aplying this to create a target doesn't seam to be practicle.
but I will create an equalization with my flat envelope method with your data so we can see what the resulting curve in var is....and maybe you try it out. just sending this post seperatly

While I can sort of understand what you mean (I think), it still is not totally clear to me what the big advantage is in not smoothing graphs when equalizing real "normal" in-room measurements -- except for the bass. Anechoic or even just quasi-anechoic gated measurements can be reasonably left unsmoothed, for sure.

I will create an equalization with my flat envelope method with your data so we can see what the resulting curve in var is....and maybe you try it out. just sending this post seperatly

From the measurements, it seems to me that besides expecting a more gradually increasing bass response, one could probably just leave the mid and high frequency band relatively flat "as-is" without further fussing about, and only apply a general ~2 kHz high shelf filter tone control when desired according to taste.

 

[ernestcarl]

One would only need two lines in the text file to create that extremely simplified curve.

Tried creating a simple 0.8dB target, but apparently one needs to add at least five lines.

I also made a few more variations of the Toole Trained Listeners curve.

Frequency Dependent Window 15 cycles

The simple 0.8 dB slope diagonal line target is probably too steep at my two meters distance. I think the "flat" HF version of the Toole trained listener's curve (trace #8) makes more sense as a simple guide since in REW's EQ window one can easily make adjustments to the HF as well as bass.

 

[ernestcarl]

I just finished watching another great Audioholics discussion with Anthony Grimani on board and I think it's really worth re-emphasizing what he says here about room target or house curves:

Benefits of Active Speakers & Digital EQ w Anthony Grimani​

Time starts around 54 min

 

[Sonic icons]

Your Toole curve is based on the typically reflective room curve. But there’s also the curve mentioned in this post. It rolls of more on the high end. Shouldn’t this actually be the Toole target curve?

I have a few comments about the Toole room curve you mentioned (I hope this old discussion is still of interest!) This looks like one of the most "objective" target curves because it is entirely measurement-based.

"The idealized steady-state room curve measured using subjectively highly-rated loudspeakers in typically reflective rooms. All [the loudspeakers that contributed to defining this particular target curve] had very flat and smooth anechoic on-axis and listening-window frequency responses."

So how about that roll-off on the high end, specifically the steep "extra" 3 dB roll-off between 10 and 20 kHz? I have a simple explanation: the tweeters roll off in this manner in most of the "highly rated speakers" that were used to create this target curve. We can expect that Revel speakers are a major contingent of the highly rated speakers. Go to the reviews on this site of Revel speakers. The top octave tweeter roll-off is clearly seen in all the graphs. Just look at the difference between the "red line" and the actual response curve above 10 kHz in the "Estimated In-Room Response". You can even see differences between the tweeter response of different Revel series. In the lower-cost Concerta2 series, such as M16 and F35, the sharp roll-off starts at 11 kHz. In Performa and PerformaBe models such as F208 and F328Be, the roll-off starts at a higher frequency.

Thus, I would reject the top octave roll-off as not a fundamental part of the target curve, since it is due to performance of a specific component, not to the speaker-room interaction (however, you may want to keep the roll-off if you know the tweeters in your speakers do this). To obtain a complete 20 Hz - 20 kHz curve, I would also extrapolate the linear rise on the low-frequency end from 60 Hz to 20 Hz. With those two edits, this target curve shows a downward tilt of -4.6 dB from 20 Hz to 20 kHz. That is less of a full-range tilt than most target curves, but because of its objective nature this target curve differs from other Harman curves in not containing a deep bass low-pass "shelf" (with cut-on frequency of about 100 Hz). You could always add a deep bass shelf if that is your preference. With added shelf height of 3 to 5 dB, the total downward tilt of the target curve would fall between 7.6 and 9.6 dB, similar to other Harman curves.

 

[FrancisCote37]

Just share my new discovery. I'm working on it to improved it.


0 2.49145
5 2.49145
9 2.49145
16 2.49145
24.9843 2.99563
49.9563 1.98014
1000.54 -0.0357277
5995.12 -1.02104
11997.2 -2.01079
22595.2 -2.96727

 

[FrancisCote37]

Another one that I like to share with you.
NAME Unnamed DEVICENAME BREAKPOINTS 0 2.5 20.0916 1.96016 50 2 325 4 700 2 1000 0 2000 0 4000 -2 8000 -1 17500 0 24000 0 LOWLIMITHZ 17.1096 HIGHLIMITHZ 24000

 

[FrancisCote37]

Another one for the interest.

NAME Unnamed DEVICENAME BREAKPOINTS 0 0 5 0 8 0 12.9544 0.0221908 159.986 0.00595305 180.138 0.997084 249.988 1.0009 399.883 -0.0100201 613.065 -1.006 705.626 -2.02667 1002.59 -3.00542 4004.87 -3.00393 6998.31 -1.97604 13495.6 -1.00062 20860 0.000737421 LOWLIMITHZ 17.1096 HIGHLIMITHZ 24000

 

[Keith_W]

I hope the OP doesn't mind, but I modified his diagram with a couple of extra target curves.

I think I must be losing my mind or my hearing. Because my preferred target curve is: none of these.

Up till recently, I have only been performing DSP correction up to the Schroder frequency. After a bit of discussion with Dr. Uli Brueggemann (author of Acourate) and a bit of reading, it turns out that a lot of effort has been made to avoid deleterious effects of unwanted full range correction on speakers via psychoacoustic filtering, complex smoothing, and so on. Since it costs me nothing to implement a full range correction, I decided to do so.

These are my listening impressions:

- Flat downward tilt. I made these at a range of slopes: -2dB, -4dB, and -6dB. The -6dB I did not like at all, it sounded too thick, had too little top end, and waaaaay too much bass. The -2dB sounded too bright (as you would expect, because it is almost a flat line). The -4dB was a happy medium, and it sounded OK ... but still too much bass.

- B&K. This was just unpleasant on my system. Sounded too bottom heavy.

- Harman / HATS: grouped them both together because they are similar. Harman has more bass impact and a real physicality to it, but certain instruments (notably piano and cello) sounded unbalanced with an overly exaggerated bottom end. HATS has no top end sparkle and sounded dull. Great bass though.

- Toole / Olive: probably ranked 2nd in terms of preference, behind the straight -4dB flat downward tilt. But still too much bass.

So, by the end of the above experiment, I was dumbfounded. WHY is it that a straight downward tilt of -2dB sound too bright, but a downward tilt of -4dB have too much bass? This lead to more iterations of the Toole/Olive filter (shown in green in the above graph) where I subjectively adjusted the amount of bass and treble to my liking.

Much to my shock and horror, I ended up with this as the best sounding curve:

This is the inverse of every known curve on the planet and all published research! Anybody who looks at this curve will think that the result should sound like a tin can! But it does not - to my ears, it sounds balanced, smooth, and with the right amount of bass to give some weight to instruments but without sounding heavy or thick. As my wife said (about some of the other target curves) - "it sounds as if the pianist was playing with the pedal all the time". I would agree. With this curve, it does not - piano notes are crisp and have a satisfying weight.

So this made me wonder whether my microphone or mic preamp was faulty and cutting off the high end (which would lead to this subjectively preferred target curve). I pulled out my Behringer ECM8000 and it gave the exact same measurement as my Earthworks M30. I ran a calibration through the RME interface - it was flawless.

Could there be a problem with Acourate? I took a verification measurement with REW:

I kept the graphs at the same scale to make them easier to compare. As you can see, they look the same.

Obviously, the last variable in this question is whether there could be something wrong with me. Has my idea of "good sound" changed so much that I now prefer such an unbalanced curve? To be honest, I am not sure. I pulled out my DCA Stealth headphones (which are known to comply with the Harman curve for headphones), and I still think they sound dull and lacking in the top end. So perhaps the problem really is with me - maybe hearing loss, maybe I am an outlier when it comes to taste. I don't know.

Anyway, as with all new filters, I need to listen more extensively and throw different types of music at it to see if it really works. These filters were generated last night, and I had a 2 hour listening session before it was lights off for me. But so far, it really does seem to work. I have a strong bias for me to not want this curve to work. I would rather think that this preference was due to the selection of music I threw at it rather than the more frightening prospect that I may be developing high freq hearing loss, which would require a massive treble boost to restore natural sounding sound.

I have asked a friend to implement the same curve on his system, and I will be going over in the next few days to find out if it works on his speakers, or whether it makes his system sound like a tin can. And I will be soliciting some opinions as to whether other people think this curve on my system sounds balanced or not. I will think about visiting an audiologist

I do not think I am some kind of audio genius who has stumbled across a new target curve that has been missed by all the leading researchers in this field. It is far more likely that there is something very, very wrong with my system or with me.

Also, those of you who want to replicate the target curve (and who are using Acourate), inbox me with your email address and I will be happy to send my Target.tgt file to you.

 

[dasdoing]

the Haarman-Tool one should be renamed to something more resambling where it comes from:

 

[Soniclife]

I hope the OP doesn't mind, but I modified his diagram with a couple of extra target curves.

I think I must be losing my mind or my hearing. Because my preferred target curve is: none of these.

View attachment 302597

Up till recently, I have only been performing DSP correction up to the Schroder frequency. After a bit of discussion with Dr. Uli Brueggemann (author of Acourate) and a bit of reading, it turns out that a lot of effort has been made to avoid deleterious effects of unwanted full range correction on speakers via psychoacoustic filtering, complex smoothing, and so on. Since it costs me nothing to implement a full range correction, I decided to do so.

These are my listening impressions:

- Flat downward tilt. I made these at a range of slopes: -2dB, -4dB, and -6dB. The -6dB I did not like at all, it sounded too thick, had too little top end, and waaaaay too much bass. The -2dB sounded too bright (as you would expect, because it is almost a flat line). The -4dB was a happy medium, and it sounded OK ... but still too much bass.

- B&K. This was just unpleasant on my system. Sounded too bottom heavy.

- Harman / HATS: grouped them both together because they are similar. Harman has more bass impact and a real physicality to it, but certain instruments (notably piano and cello) sounded unbalanced with an overly exaggerated bottom end. HATS has no top end sparkle and sounded dull. Great bass though.

- Toole / Olive: probably ranked 2nd in terms of preference, behind the straight -4dB flat downward tilt. But still too much bass.

So, by the end of the above experiment, I was dumbfounded. WHY is it that a straight downward tilt of -2dB sound too bright, but a downward tilt of -4dB have too much bass? This lead to more iterations of the Toole/Olive filter (shown in green in the above graph) where I subjectively adjusted the amount of bass and treble to my liking.

Much to my shock and horror, I ended up with this as the best sounding curve:

This is the inverse of every known curve on the planet and all published research! Anybody who looks at this curve will think that the result should sound like a tin can! But it does not - to my ears, it sounds balanced, smooth, and with the right amount of bass to give some weight to instruments but without sounding heavy or thick. As my wife said (about some of the other target curves) - "it sounds as if the pianist was playing with the pedal all the time". I would agree. With this curve, it does not - piano notes are crisp and have a satisfying weight.

So this made me wonder whether my microphone or mic preamp was faulty and cutting off the high end (which would lead to this subjectively preferred target curve). I pulled out my Behringer ECM8000 and it gave the exact same measurement as my Earthworks M30. I ran a calibration through the RME interface - it was flawless.

Could there be a problem with Acourate? I took a verification measurement with REW:

I kept the graphs at the same scale to make them easier to compare. As you can see, they look the same.

Obviously, the last variable in this question is whether there could be something wrong with me. Has my idea of "good sound" changed so much that I now prefer such an unbalanced curve? To be honest, I am not sure. I pulled out my DCA Stealth headphones (which are known to comply with the Harman curve for headphones), and I still think they sound dull and lacking in the top end. So perhaps the problem really is with me - maybe hearing loss, maybe I am an outlier when it comes to taste. I don't know.

Anyway, as with all new filters, I need to listen more extensively and throw different types of music at it to see if it really works. These filters were generated last night, and I had a 2 hour listening session before it was lights off for me. But so far, it really does seem to work. I have a strong bias for me to not want this curve to work. I would rather think that this preference was due to the selection of music I threw at it rather than the more frightening prospect that I may be developing high freq hearing loss, which would require a massive treble boost to restore natural sounding sound.

I have asked a friend to implement the same curve on his system, and I will be going over in the next few days to find out if it works on his speakers, or whether it makes his system sound like a tin can. And I will be soliciting some opinions as to whether other people think this curve on my system sounds balanced or not. I will think about visiting an audiologist

I do not think I am some kind of audio genius who has stumbled across a new target curve that has been missed by all the leading researchers in this field. It is far more likely that there is something very, very wrong with my system or with me.

Also, those of you who want to replicate the target curve (and who are using Acourate), inbox me with your email address and I will be happy to send my Target.tgt file to you.

I've had similar experiences, what should be horrible seeming to be great, after about a week reality seems to assert itself and you suddenly think, agh, that's horrible. So keep listening and enjoy it, in time you will find out if it's great or not. I like to put my library on shuffle for this type of listening, not knowing what's just started playing seems to give me a fresh perspective without baggage.

 

[ernestcarl]

I hope the OP doesn't mind, but I modified his diagram with a couple of extra target curves.

I think I must be losing my mind or my hearing. Because my preferred target curve is: none of these.

View attachment 302597

Up till recently, I have only been performing DSP correction up to the Schroder frequency. After a bit of discussion with Dr. Uli Brueggemann (author of Acourate) and a bit of reading, it turns out that a lot of effort has been made to avoid deleterious effects of unwanted full range correction on speakers via psychoacoustic filtering, complex smoothing, and so on. Since it costs me nothing to implement a full range correction, I decided to do so.

These are my listening impressions:

- Flat downward tilt. I made these at a range of slopes: -2dB, -4dB, and -6dB. The -6dB I did not like at all, it sounded too thick, had too little top end, and waaaaay too much bass. The -2dB sounded too bright (as you would expect, because it is almost a flat line). The -4dB was a happy medium, and it sounded OK ... but still too much bass.

- B&K. This was just unpleasant on my system. Sounded too bottom heavy.

- Harman / HATS: grouped them both together because they are similar. Harman has more bass impact and a real physicality to it, but certain instruments (notably piano and cello) sounded unbalanced with an overly exaggerated bottom end. HATS has no top end sparkle and sounded dull. Great bass though.

- Toole / Olive: probably ranked 2nd in terms of preference, behind the straight -4dB flat downward tilt. But still too much bass.

So, by the end of the above experiment, I was dumbfounded. WHY is it that a straight downward tilt of -2dB sound too bright, but a downward tilt of -4dB have too much bass? This lead to more iterations of the Toole/Olive filter (shown in green in the above graph) where I subjectively adjusted the amount of bass and treble to my liking.

Much to my shock and horror, I ended up with this as the best sounding curve:

This is the inverse of every known curve on the planet and all published research! Anybody who looks at this curve will think that the result should sound like a tin can! But it does not - to my ears, it sounds balanced, smooth, and with the right amount of bass to give some weight to instruments but without sounding heavy or thick. As my wife said (about some of the other target curves) - "it sounds as if the pianist was playing with the pedal all the time". I would agree. With this curve, it does not - piano notes are crisp and have a satisfying weight.

So this made me wonder whether my microphone or mic preamp was faulty and cutting off the high end (which would lead to this subjectively preferred target curve). I pulled out my Behringer ECM8000 and it gave the exact same measurement as my Earthworks M30. I ran a calibration through the RME interface - it was flawless.

Could there be a problem with Acourate? I took a verification measurement with REW:

I kept the graphs at the same scale to make them easier to compare. As you can see, they look the same.

Obviously, the last variable in this question is whether there could be something wrong with me. Has my idea of "good sound" changed so much that I now prefer such an unbalanced curve? To be honest, I am not sure. I pulled out my DCA Stealth headphones (which are known to comply with the Harman curve for headphones), and I still think they sound dull and lacking in the top end. So perhaps the problem really is with me - maybe hearing loss, maybe I am an outlier when it comes to taste. I don't know.

Anyway, as with all new filters, I need to listen more extensively and throw different types of music at it to see if it really works. These filters were generated last night, and I had a 2 hour listening session before it was lights off for me. But so far, it really does seem to work. I have a strong bias for me to not want this curve to work. I would rather think that this preference was due to the selection of music I threw at it rather than the more frightening prospect that I may be developing high freq hearing loss, which would require a massive treble boost to restore natural sounding sound.

I have asked a friend to implement the same curve on his system, and I will be going over in the next few days to find out if it works on his speakers, or whether it makes his system sound like a tin can. And I will be soliciting some opinions as to whether other people think this curve on my system sounds balanced or not. I will think about visiting an audiologist

I do not think I am some kind of audio genius who has stumbled across a new target curve that has been missed by all the leading researchers in this field. It is far more likely that there is something very, very wrong with my system or with me.

Also, those of you who want to replicate the target curve (and who are using Acourate), inbox me with your email address and I will be happy to send my Target.tgt file to you.

This is rather an extreme case...

If you don't mind, could you post the mdat file of the corrected measurements shown?

 

[Bach]

Let us say you are happy with the Harman correction of 5dB below 100 Hz. Well, do not be surprised that you cannot play as loud anymore and watch out that the woofer cone does not get ejectted into your room; assuming you have a speaker with a 5 to 7 inch woofer and bass reflex port tuned above 40Hz. Also keep in mind that the distortion in the bass region will also be very much more at higher listening levels.
So, i wish everone a lot of wishdom at augmenting the bass level of their two way bookshelf speakers with big WAF, which of course is contrary to big WOOF (ers).