Best Room Response

[JR4321]

Is it possible that an un-optimized spectrum is the preferred response for your room? I've experimented with maybe 2 dozen custom targets and corresponding filters with Dirac.

I keep coming back to the un-optimized response being my preferred.

So theory...is it possible that your natural room's response will sound the most "real" with your system? Any alterations from the unaltered response seem to sound unnatural to me... Just a thought.

My room:

 

[richard12511]

Is it possible that an un-optimized spectrum is the preferred response for your room? I've experimented with maybe 2 dozen custom targets and corresponding filters with Dirac.

I keep coming back to the un-optimized response being my preferred.

So theory...is it possible that your natural room's response will sound the most "real" with your system? Any alterations from the unaltered response seem to sound unnatural to me... Just a thought.

My room:

It's possible, but man, that doesn't look great. Have tried a Harman curve instead of the default Dirac curve? That's what I use, and I prefer it to the Dirac default curve.

 

[JR4321]

It's possible, but man, that doesn't look great. Have tried a Harman curve instead of the default Dirac curve? That's what I use, and I prefer it to the Dirac default curve.

I haven't. Just looked it up. I'll give that a try. Thanks!

 

[JR4321]

I haven't. Just looked it up. I'll give that a try. Thanks!

Actually, what I found was his ideal Headphone response. Does he have an ideal speaker response?

 

[richard12511]

This is the one I use. The "harman.txt" file is with "0" being the max of the graph, so there will be no (or very little)boosting. You may want to move the overall curve up or down to more closely track through the center of your pre-eq curve. The closer the curve tracks to your pre-eq curve, the less Dirac will have to do, though you'll probably end up with more boosts. Just make sure you add an equal amount to every point, as that way you'll preserve the shape of the curve. The "+4" curve I also attached is an example to show you what I mean. I just added "4" to all the values.

Also you may want to try only EQing up to a certain point, but you'll need add to the curve(like I described above) to make sure the point at which you stop correction isn't too far above or below the line. For my system, I do have a full range EQ setup that sounds great, but for 2 channel music, I slightly prefer the setup where I only EQ to 700hz. Looking at your measurements, your speakers seem to have a dip from 500-1.8k. If you're not gonna do full range EQ, I would experiment with stopping the EQ both just before that dip as well as just after that dip to see which you prefer.

What speakers are these?

 

[JR4321]

This is the one I use. The "harman.txt" file is with "0" being the max of the graph, so there will be no (or very little)boosting. You may want to move the overall curve up or down to more closely track through the center of your pre-eq curve. The closer the curve tracks to your pre-eq curve, the less Dirac will have to do, though you'll probably end up with more boosts. Just make sure you add an equal amount to every point, as that way you'll preserve the shape of the curve. The "+4" curve I also attached is an example to show you what I mean. I just added "4" to all the values.

Also you may want to try only EQing up to a certain point, but you'll need add to the curve(like I described above) to make sure the point at which you stop correction isn't too far above or below the line. For my system, I do have a full range EQ setup that sounds great, but for 2 channel music, I slightly prefer the setup where I only EQ to 700hz. Looking at your measurements, your speakers seem to have a dip from 500-1.8k. If you're not gonna do full range EQ, I would experiment with stopping the EQ both just before that dip as well as just after that dip to see which you prefer.

What speakers are these?

Thanks for the info. Your curve is close to one of the more common I've used. I still use Dirac 1.X Mac-based software (not Live), FYI.

Speakers are Magnepan 3.7i, with a pair of Rel Strata subs. Room is 24' X 14', with speakers 10' from the front wall.

 

[JR4321]

Thanks for the info. Your curve is close to one of the more common I've used. I still use Dirac 1.X Mac-based software (not Live), FYI.

Speakers are Magnepan 3.7i, with a pair of Rel Strata subs. Room is 24' X 14', with speakers 10' from the front wall.

One example of an optimized curve. This gives a slight "cave-like" sound (like the singer is coming from inside a cave). Way less natural sounding vs the un-optimized.

 

[thewas]

Correcting above the modal and transition region to a predefined target curve can be problematic as at the listeners position you measure a mix of both direct and reflected sound, but to our perception a flat direct sound is what sounds neutral, so you need to have anechoic (directivity) measurements of your loudspeakers. See more here https://www.audiosciencereview.com/...ut-room-curve-targets-room-eq-and-more.10950/

 

[JR4321]

Correcting above the modal and transition region to a predefined target curve can be problematic as at the listeners position you measure a mix of both direct and reflected sound, but to our perception a flat direct sound is what sounds neutral, so you need to have anechoic (directivity) measurements of your loudspeakers. See more here https://www.audiosciencereview.com/...ut-room-curve-targets-room-eq-and-more.10950/

Great info; thanks.

More testing is revealed to me that the mic angle makes a significant difference. Room correction with the mic vertical vs horizontal gives much different results, though the corrected profiles are nearly identical.

 

[andreasmaaan]

@JR4321, I would suggest manually equalising your in-room response only below around 300 or 400Hz.

Above this frequency range, our brains are much better at separating the direct sound from the speakers from the reflections, and it is mostly the direct sound that we perceive as the tonal balance of the system. It is therefore better not to EQ to compensate for the in-room response in this range.

What you may wish to try, however, is taking gated measurements of your loudspeakers and then correcting their quasi-anechoic response in this mid-high frequency range.

IME (and in my view the science supports this), room EQ in the lower frequencies and loudspeaker EQ (based on gated measurements) in the higher frequencies is the best approach to using EQ to correct your system.

However, for loudspeaker correction, you'd really want to be able to trust your measurement mic. Also, you'd want to take a number of measurements across a +/-30° (or so) listening window, with your correction curve taking into account both the on-axis response and an average of this "listening window" response (even better would be full gated polar measurements, but this may arguably be getting into the realm of diminishing returns, especially if your speakers are of a conventional type and therefore have a polar response that is relatively easy to predict).

Out of interest, what speakers do you have?

EDIT: ah, @thewas_ beat me to it

 

[JR4321]

@JR4321, I would suggest manually equalising your in-room response only below around 300 or 400Hz.

Above this frequency range, our brains are much better at separating the direct sound from the speakers from the reflections, and it is mostly the direct sound that we perceive as the tonal balance of the system. It is therefore better not to EQ to compensate for the in-room response in this range.

What you may wish to try, however, is taking gated measurements of your loudspeakers and then correcting their quasi-anechoic response in this mid-high frequency range.

IME (and in my view the science supports this), room EQ in the lower frequencies and loudspeaker EQ (based on gated measurements) in the higher frequencies is the best approach to using EQ to correct your system.

However, for loudspeaker correction, you'd really want to be able to trust your measurement mic. Also, you'd want to take a number of measurements across a +/-30° (or so) listening window, with your correction curve taking into account both the on-axis response and an average of this "listening window" response (even better would be full gated polar measurements, but this may arguably be getting into the realm of diminishing returns, especially if your speakers are of a conventional type and therefore have a polar response that is relatively easy to predict).

Out of interest, what speakers do you have?

EDIT: ah, @thewas_ beat me to it

Can't say I followed all of that: gated measurement, room eq vs speaker eq... but I'll do some Googling. I think you're talking about room treatments like dampening and dispersion vs the Dirac, but possibly not.

I think I'm going to play around with sub and speaker positioning next. I feel like I've exhausted the possibilities of Dirac for the moment and aren't having great results.

 

[andreasmaaan]

Can't say I followed all of that: gated measurement, room eq vs speaker eq... but I'll do some Googling. I think you're talking about room treatments like dampening and dispersion vs the Dirac, but possibly not.

Actually no, not room treatment.

I was talking only about using EQ to correct your system, but what I was saying was that you should either:

1. correct only below 300-400Hz (i.e. correct only for the room) or
2. correct the room below 300-400Hz and correct the speakers above 300-400Hz

Option 1 requires only a basic measurement mic and a bit of reading and working things out.

Option 2 is a lot more involved, and requires taking gated measurements of your speakers from a variety of different angles. It would take a fair bit more research and learning on your part.

Whether 2 is worthwhile to you depends firstly on your speakers (the better the speakers are out of the box, the less there is to be gained), and secondly on your level of interest in learning about loudspeaker measurement, and actually to some extent loudspeaker design.

So I'm definitely not suggesting you try option 2, just putting it out there...

Anyway, the essential point was the same as @thewas_'s: for room correction, you shouldn't change anything higher in frequency than about 300-400Hz.

 

[QMuse]

Actually no, not room treatment.

I was talking only about using EQ to correct your system, but what I was saying was that you should either:

1. correct only below 300-400Hz (i.e. correct only for the room) or
2. correct the room below 300-400Hz and correct the speakers above 300-400Hz

Option 1 requires only a basic measurement mic and a bit of reading and working things out.

Option 2 is a lot more involved, and requires taking gated measurements of your speakers from a variety of different angles. It would take a fair bit more research and learning on your part.

Whether 2 is worthwhile to you depends firstly on your speakers (the better the speakers are out of the box, the less there is to be gained), and secondly on your level of interest in learning about loudspeaker measurement, and actually to some extent loudspeaker design.

So I'm definitely not suggesting you try option 2, just putting it out there...

Anyway, the essential point was the same as @thewas_'s: for room correction, you shouldn't change anything higher in frequency than about 300-400Hz.

I would mostly agree with this, except for one detail that @thewas_ mentioned but you didn't, and that is transition region. IMO opinion room effects doesn't magically stop like hitting a wall at 400Hz - physics simply doesn't work that way. What I believe is happening is that room's influence on loudspeaker response slowly diminishes but still extending to app 800Hz, and that region (400-800Hz) is considered as "transition" region. IMO room EQ can and should be used in that region but with smaller scale gain corrections and lower Q values. After app 800Hz room is leaving the scene and, as you said, we are left to correcting a loudspeaker using pseudo-anechoic measurements.

 

[andreasmaaan]

I would mostly agree with this, except for one detail that @thewas_ mentioned but you didn't, and that is transition region. IMO opinion room effects doesn't magically stop like hitting a wall at 400Hz - physics simply doesn't work that way. What I believe is happening is that room's influence on loudspeaker response slowly diminishes but still extending to app 800Hz, and that region (400-800Hz) is considered as "transition" region. IMO room EQ can and should be used in that region but with smaller scale gain corrections and lower Q values. After app 800Hz room is leaving the scene and, as you said, we are left to correcting a loudspeaker using pseudo-anechoic measurements.

I totally agree, which is why I gave a range of 300-400Hz (although it's good that you explained it in more depth).

And all these figures we're talking about are room-dependent too ofc: in general, the larger/deader the room, the lower in frequency the transition region will be.

Having said that, as a rule of thumb, in most circumstances I would not suggest correcting above 400-ish Hz (if even that high), since wavelengths above this frequency start to get quite short in relation to likely changes in listening position (even for a single listener moving back-forward or side-side in their listening chair).

 

[Mnyb]

Also OP has panel speakers ,they can beam a lot in higher frequency so room inteaction is less anyway , they are more like giant in room headphones in some way they reflect to but probably even more delayd than typical speakers ? can they not even be cardoid in some part of the range ie barely have direct side reflections and barely direct roof or floor reflection due to line source effects . Inderect sound gets to the sides and roof/floor eventually .

 

[Willem]

since wavelengths above this frequency start to get quite short in relation to likely changes in listening position (even for a single listener moving back-forward or side-side in their listening chair).

Indeed. It is important to realize that at higher frequencies the correctly equalized listening position gets narrower and narrower. So in small rooms with a much higher transition frequency, equalization is less effective because the problems are not limited to the lowest frequencies, but also exist at higher frequencies where they are both more obnoxious and harder to equalize. Since multiple subs give a smoother response over a much larger listening area equalization is far more useful with multiple subs than with a single sub, unless you don't mind being stuck all alone in your one and only chair.
So my practical advice for small rooms is always not to bother with subs or equalization, and use bookshelf speakers with limited but clean bass extension. This also saves domestic space. For larger rooms subs are great, but only if you use at least two, with dsp equalization of one kind or another.

 

[bluefuzz]

One example of an optimized curve. This gives a slight "cave-like" sound (like the singer is coming from inside a cave). Way less natural sounding vs the un-optimized.

I'm a fairly recent convert to Dirac, using the full frequency version of Dirac Live and am generally very pleased with it. I think I know what you mean by the 'cave like' effect, however unlike you I rather like it. It is a consequence of the sound stage being much more precise and well defined. At least in my room I experience much more 'depth' to the soundstage as well as better left to right spatial precision which in some recordings make the center vocals seem farther away. I suppose it is what some audiophiles would call more 'analytical' which some people don't like.

I use the small 'chair' calibration setting with the listening position centred in a roughly 60 cm cube for the other mic positions. My listening position is pretty static and I don't really care what it sounds like elsewhere in the room.

Contrary to some of the recommendations in this thread I have not had good results correcting only below the room transition frequency. But I have had universally excellent results using the full frequency correction on three wildly different sets of speakers. I have tried Dirac on my Linkwitz LXminis, a pair of heavily modded vintage JBL L100s (rebuilt with symmetrical driver layout, copious internal bracing and damping with Troels Gravesens crossover and an aperiodic vent) plus a pair of small DIY towers with 5"/1" Vifa drivers in an MTM configuration. I have one 10" DIY subwoofer with passive radiator. In all cases I have found the full frequency Dirac to be a massive improvement to the frequency response, perceived accuracy, soundstage – well everything really! I use more or less the default Dirac/Harman curve ...

 

[JR4321]

Actually no, not room treatment.

I was talking only about using EQ to correct your system, but what I was saying was that you should either:

1. correct only below 300-400Hz (i.e. correct only for the room) or
2. correct the room below 300-400Hz and correct the speakers above 300-400Hz

Option 1 requires only a basic measurement mic and a bit of reading and working things out.

Option 2 is a lot more involved, and requires taking gated measurements of your speakers from a variety of different angles. It would take a fair bit more research and learning on your part.

Whether 2 is worthwhile to you depends firstly on your speakers (the better the speakers are out of the box, the less there is to be gained), and secondly on your level of interest in learning about loudspeaker measurement, and actually to some extent loudspeaker design.

So I'm definitely not suggesting you try option 2, just putting it out there...

Anyway, the essential point was the same as @thewas_'s: for room correction, you shouldn't change anything higher in frequency than about 300-400Hz.

OK, thanks for the clarification. I have experimented with shifting the frequencies EQ'd, and that does solve the "cave-like" sound I mentioned.

What would be very interesting/helpful would be some kind of guide as to what frequencies or phase adjustments cause what effects. For example, I know a phase change can affect the apparent width of the soundstage e.g. Carver's Sonic Hologram - C9: https://www.audiosciencereview.com/...-line-source-speakers.1261/page-3#post-362388 Though I don't think I can adjust this with Dirac.

I still haven't determined what causes the "cave-like" effect I described earlier. It must be in the 400-1000Hz range, as it is most noticeable in the male vocals, but changing my filters for this region hasn't rid me of it. I'll keep experimenting.

 

[JR4321]

I'm a fairly recent convert to Dirac, using the full frequency version of Dirac Live and am generally very pleased with it. I think I know what you mean by the 'cave like' effect, however unlike you I rather like it. It is a consequence of the sound stage being much more precise and well defined. At least in my room I experience much more 'depth' to the soundstage as well as better left to right spatial precision which in some recordings make the center vocals seem farther away. I suppose it is what some audiophiles would call more 'analytical' which some people don't like.

I use the small 'chair' calibration setting with the listening position centred in a roughly 60 cm cube for the other mic positions. My listening position is pretty static and I don't really care what it sounds like elsewhere in the room.

Contrary to some of the recommendations in this thread I have not had good results correcting only below the room transition frequency. But I have had universally excellent results using the full frequency correction on three wildly different sets of speakers. I have tried Dirac on my Linkwitz LXminis, a pair of heavily modded vintage JBL L100s (rebuilt with symmetrical driver layout, copious internal bracing and damping with Troels Gravesens crossover and an aperiodic vent) plus a pair of small DIY towers with 5"/1" Vifa drivers in an MTM configuration. I have one 10" DIY subwoofer with passive radiator. In all cases I have found the full frequency Dirac to be a massive improvement to the frequency response, perceived accuracy, soundstage – well everything really! I use more or less the default Dirac/Harman curve ...

Thanks for sharing your experiences with Dirac. I originally loved this software, but for whatever reason, I can't seem to get it to work as I'd like these days. I'll keep playing around with it. I have found the mic angle to be very sensitive. Also, whether I'm sitting in the listening position or just the mic seems to make a noticeable difference.

In the end, I'm not sure how accurate the measurements are. So it seems it's just a matter of finding a profile/setup that sounds best to me.

 

[QMuse]

Indeed. It is important to realize that at higher frequencies the correctly equalized listening position gets narrower and narrower.

That is not my experience. I did some mild EQ in the 400-800Hz range and it is valid for both my LPs in the 45m2 room. The same is of course true for the speaker's correction in 800-20kHz range where I did some minor corrections based on pseudo anechoic gated measurements.

At my 2nd LP I only had to adapt 20-400Hz range and even there I pretty much only had to deal with room mode at 58Hz which was causing a huge peak at 2nd LP but not so at my primary LP.

Why do you think correction in 400-800Hz range is not valid throughout the room when room influence there is actually much less then below Schroeder?

EDIT: Here is my left channel filter for primary LP (blue, 4m from the speakers) and 2nd LP (red, 6m from speakers). As you can see 400-800Hz range is practically identical. it's pretty much the same thing with right channel.