What is your favorite house curve

[Rednaxela]

I always end up with about 4dB droop from 100Hz to 10kHz. A little more is generally OK, but things can get kind of fuzzy.

So that’s about 0.6dB/oct. I have just dialed in a straight house ‘curve’ with that slope, and first impressions are good. It’s my favourite so far in fact.

 

[Rednaxela]

This morning I went back to B&K, with the house curve values found here.

Aligning it a bit higher for a little bass boost, and only partially pulling down the mid range bump, I got this, which doesn't sound bad at all.

Still looking for a more informed way of dealing with the HF region. But that's probably something for another thread.

 

[digitalfrost]

After getting into IEMs I wanted my stereo and IEMs sound closer to each other and I ended up with a curve that has a slight bass boost and is only 3dB down at 20khz. Note I also changed the listening spot and re-adjusted by subwoofer levels pre-correction to get them more inline with the speakers.

My point is other people's target curves may or may not fit your current situation.

Also one might think the 0.3dB bass boost would be negligible but it makes all the difference.

0 -40.0
28 -20.0
29 0.3
35 0.3
41 0.3
47 0.29
53 0.29
59 0.28
64 0.28
70 0.27
76 0.26
82 0.25
88 0.23
94 0.22
100 0.21
105 0.19
1000 0.0
20000 -3.0
22000 -40.0
24000 -100.0

 

[Robbo99999]

This morning I went back to B&K, with the house curve values found here.

Aligning it a bit higher for a little bass boost, and only partially pulling down the mid range bump, I got this, which doesn't sound bad at all.

View attachment 237351

Still looking for a more informed way of dealing with the HF region. But that's probably something for another thread.

Ideal situation is likely having your model of speaker measured by Amir on his Klippel (or perhaps it's one he's already measured), which then allow you to do an Anechoic EQ of the speaker to make it Anechoic Flat - that would sort out the frequency response from around 400-20000Hz, then you'd do roomEQ on the bass where the room dominates the behaviour. You'd probably EQ the bass to something similar to the Harman Curve. So you'd ignore in-room measurements above 400Hz and would rely on the Anechoic EQ there.

 

[Rednaxela]

Thanks @Robbo99999.

So I’ve been thinking along the same lines. In fact I have some quasi-anechoic data myself. Strictly anechoic flat is going to be difficult I’m afraid. Which doesn’t mean there’s nothing I can do at all. But it will be a bit more of a balancing act maybe.

 

[Robbo99999]

Thanks @Robbo99999.

So I’ve been thinking along the same lines. In fact I have some quasi-anechoic data myself. Strictly anechoic flat is going to be difficult I’m afraid. Which doesn’t mean there’s nothing I can do at all. But it will be a bit more of a balancing act maybe.

Yeah, you'll have to weigh up how valid your captures are, and then go from there I suppose.

 

[Rednaxela]

Ok, so what I just did was I averaged all measurements of one speaker, on and off-axis.

It took two bands to correct that to flat (enough). These two bands I applied to my in-room measurement. This straightened out the predicted HF response very nicely. I then aligned the BK target to this predicted HF curve. Interestingly this aligned very well with the bottom end of the in-room measurement too. This left me with eight bands to iron out all the remaining kinks from the transition zone down.

The end result is very similar to this one, the biggest difference being that cut filter 9 was replaced by a broad and gentle boost around 4kHz. A very long road to a rather small change but what a difference it makes. Strings sound full and lively, and while I couldn’t get voices right with the previous corrections, this now seems to fall into place as well. Very happy with this adjustment.

 

[onununo]

Hi folks,
I'm getting mad trying to find a house curve for my system that I can call "definitive", i.e. works well with most of the contents from old 70's rock to the most recent rap, through classical music and even movies. My system is composed of a pair of Neumann's studio monitors KH310, and a pair of their KH750 DSP subs, at the sides of a large TV, located unsimmetrically in a room of irregula shape of approx. 120m3 volume. I'm using Neumann's MA-1 calibration suite and mike but I find that the automatic calibration offered is a tad bass shy and harsh sounding. The point is I am puzzled as to what adjustments should I make to their proposed house curve (it is flat from 100 to approx. 3-5kHz, then gradually rolling off to about -3 dB at 20kHz, and raised by 2 dB in the range 20-100Hz by means of a low shelf filter with Q approx. equal to 2). The point is, that if today I am delighted by selecting a house curve say linearly dropping from 20 to 20k by 10 dB, hearing the power of the bass, then tomorrow I notice that speech in movies sounds too dark, so I completely change my mind and go for a much flatter curve, but then again, I would like more presence when listening to 70's rock... It never ends!!!!
So dowmnto my own 2 cents on this subject: having speakers with a reasonably flat response in anecoic conditions, and with decent dispersion characteristics, I will NOT equalize the satellites. They really sound better if not equalized at all. What I will do is to equalize only the subs, or say, the spectral region from 18 to 100-200 Hz at most, and try (I repeat: "try") to find a matching level.

 

[nathan]

Welcome to the wonderful world of different recordings being made under different conditions and with different goals, cultural standards, and ears at the helm.

There is no way that a mono Beetles track mixed and mastered in 1966 (designed to have less bass so that the needle on a portable record player wouldn't jump out of the groove, and to allow room in the single channel to differentiate all instruments) is going to blossom with the same settings as Nolan's Batman movies on your home sound system.

This is why people as well regarded as Floyd Toole recommend that you feel free to adjust the treble and bass tone controls any time you feel a recording is lacking. This is admittedly more about two channel recordings / music recordings, which tend to be the Wild West in terms of the conditions of their creation versus film and television soundtracks which at least have some industry standards about the room, the response, the decay, the levels......even if too many production houses play fast and loose with those standards.

 

[AnalogSteph]

Trying to EQ a pair of Edifier R1280Ts in a large-ish, very lively room turned out to be quite insightful for me. If I EQ'd the bass so that it measured flat, the end result sounded more like a portable radio (no kidding), definitely not right. I then modified the room curve (bass shelf) in REW to broadly follow what the room does, minus mode eccentricities. Way better.

IMHO a room curve is an end result, not a target. Above a few hundred Hz, a measurement mic at MLP is not going to match our perception very well. I would rather take a windowed measurement in nearfield up there and merge the results, or let it be altogether. I have also found that applying some smoothing such as 1/12th oct makes resulting EQs much more palatable.

 

[JoachimStrobel]

About three years ago I started using room curves. I still find it totally “frightening” how a full frequency room curve changes the sound. It make me wonder what I could do right or wrong. The first response could be to limit the correction to 500 hz, shying away from a potential disaster. That may work with a 2ch system. But timbre matching and impulse correction in a Mch system needs a full range. On top, I believe that a Mch system following Dolby guidelines with all speakers circling the listener produces a lot of direct sound that is the source of non-natural high frequency content. Careful selection of a high frequency damping room curve is a must. Initially I avoided the ones that dampen 10khz by 4-6db as I believed that this is too much. But simply doing a quick analyses with an IPhone and AudioTools in a jazzclub that has no amplification taught me that those room curves are ok. The one in Toole’s book is good. And his remark is interesting. I remember times when the true audiophile would buy stuff without tone control which was looked at as the ketchup for non-Connaisseur. But he his of course right. A bit of treble and bass correction here and there is OK. Or a set of Dirac corrections, one for each label and engineer.

 

[Rednaxela]

Another attempt at hitting the B&K target.

This time I measured a smaller area, staying closer to the MLP. The >500Hz corrections are based on semi-anechoic data, the others on the B&K curve. Happy to see how well the post-EQ measurement tracks the REW prediction.

With music this sounds really nice. I love the improved tonal balance. Everything sounds a lot more right. Not sure about TV dialogue though, which can sometimes sound a bit veiled. Could be multiple things - will have too look further into this.

 

[luft262]

So I understand that you want the dB to slope downward as the frequency increases, but I don't see a consensus. If you dropped about 10dB from 20Hz to 10KHz would that be optimal? Obviously it's subjective, but what curve/slope would be closest to the generally most desirable house curve? How much slope to too much vs too little?

Thanks!

 

[digitalfrost]

The way the different DRC systems implement the curve it is impossible to make hard statements in terms of what is best. I have been using an 11dB slope from 20hz-20khz in the past, at the moment, all I do is a 3dB drop off from 1khz to 20khz, so barely anything.

Might be a change in taste, might be different settings. Whatever works for you. Any serious kind of DRC will impement some form of frequency dependent windowing. Depending on how you tune it, the correction will be very strong or minimal. Just find what works for you.

You cannot compare target curves across different DRC software and different systems. I have been using a straight curve in the past, but I found that it fattens up the muddy region too much, so I would always advocate to have a knee in there (like here https://www.audiosciencereview.com/...s-your-favorite-house-curve.2382/post-1349346) and if you need more bass you can always put a bass shelf below.

Also upper and lower region influence each other. There's different ways to do 10dB slopes...

Here's 2dB bass and -8dB down at 20khz


and here's 4dB bass and -6dB down at 20khz



What that will sound like depends on a lot of things you have to experiment. I would always recommend to have the knee though. I had this in the past and while it was a great improvement compared to no DRC, it can be much better than this:

 

[dlaloum]

So I understand that you want the dB to slope downward as the frequency increases, but I don't see a consensus. If you dropped about 10dB from 20Hz to 10KHz would that be optimal? Obviously it's subjective, but what curve/slope would be closest to the generally most desirable house curve? How much slope to too much vs too little?

Thanks!

Seems to me, that the sloped standard curve, is the outcome of a flat direct speaker response, with the impact of the room...

So applying a sloped target curve should in theory be an outcome rather than a "target" as such...

What if your room is more reflective? you would have more high frequencies in the mix, and you resulting profile would be flatter.

What if your room has all its walls covered by bookshelves and books (ie: absorbent) - then you would expect the profile to have a steeper slope and pitch downwards more...

In theory, I would like the "Direct" sound to be flat(ish) - but the sum of direct & reflected, in what is typically considered a "good" room (live end - dead end?) should have a slope.

When Dirac, Audyssey and others measure the F/R and show their target curves, are they measuring the sum of direct and reflected (the "voicing of the room" so to speak) - or just the direct sound from the speakers?

Is the room target curve a flawed approach? - do we need to understand how each systems target curve is defined before we can determine the optimal use of that systems target curve applications?

 

[Floyd Toole]

IMHO a room curve is an end result, not a target. Above a few hundred Hz, a measurement mic at MLP is not going to match our perception very well. I would rather take a windowed measurement in nearfield up there and merge the results, or let it be altogether. I have also found that applying some smoothing such as 1/12th oct makes resulting EQs much more palatable.

It is more than your HO, it is fact. I have not followed this thread, but it just came to my attention. It is clear that some participants have the right idea and others are still caught up in the never-ending dilemma caused by a belief that the steady-state room curve is a reliable indicator of potential sound quality. Put differently, an omni microphone is not equivalent to two ears and a brain. Things that may look alarming in "room curves" may be perceived by a binaural-hearing human as innocent, or even pleasant, spaciousness. It is the direct sound that dominates perceived timbre/sound quality, and years of double-blind testing indicates that smooth and flat is a good starting place. That is best determined in anechoic measurements of the loudspeakers or time-windowed FFTs in the room (above about 1kHz, typically). Below about 500 Hz bass can be narrow-band equalized - attenuating peaks, but ignoring narrow dips.
It is amusing that the ancient B&K target curve is still trotted out. Read the original paper and see the frequency responses of the grossly colored speakers used in their test - typical of the time (1972 I think), but nowadays nobody in this forum would give them house room - loudspeaker technology has advanced. That said, as a broad "average" curve, the tendency is correct.

The room curves associated with the best sounding conventional forward-firing loudspeakers are all downward tilted - why? Remember the curve is "an end result, not a target" and conventional loudspeakers are omnidirectional at very low frequencies, becoming more directional as frequency rises. So, if such loudspeakers are designed with desirable flat on-axis/listening window frequency responses (the first arriving "direct" sound) the total sound output decays with increasing frequency. Steady state room curves measure direct + all reflected sound so they must tilt downward - or, more accurately, rise towards the bass. Then there is air absorption, which adds more downward tilt above about 1 kHz. It is all understood and predictable. Also understood is that the detailed shape of a room curve depends on which loudspeaker is used. Even if designed to be flat on axis the off-axis performance is the dominant factor in determining the shape of the curve - equalization cannot, can not, compensate for uneven off-axis sound radiation. Only a better loudspeaker can do that. If you search my responses in this forum in recent years you will find more explanation. That is why we should be grateful to see useful technical measurements (called the spinorama) on this and other forums. Starting with a well designed loudspeaker will yield a smoothish, slightly tilted room curve at the listening position - except at low frequencies and that is a separate story, with multiple solutions. It is not a target curve, though, and adjusting flawed loudspeakers to imitate the shape will not yield equally good sound. Sorry.

Finally, there are the inevitable variations in spectral balance in recordings. This is why one needs tone controls, if one is fussy.

So, if you have found "perfection" in sound, good for you. Enjoy it. But, your room curve tells me nothing useful unless I can see a spinorama or equivalent anechoic data on the loudspeaker that produced it. Otherwise the forum discussions of what curve is right or wrong cannot end.

 

[luft262]

It is more than your HO, it is fact. I have not followed this thread, but it just came to my attention. It is clear that some participants have the right idea and others are still caught up in the never-ending dilemma caused by a belief that the steady-state room curve is a reliable indicator of potential sound quality. Put differently, an omni microphone is not equivalent to two ears and a brain. Things that may look alarming in "room curves" may be perceived by a binaural-hearing human as innocent, or even pleasant, spaciousness. It is the direct sound that dominates perceived timbre/sound quality, and years of double-blind testing indicates that smooth and flat is a good starting place. That is best determined in anechoic measurements of the loudspeakers or time-windowed FFTs in the room (above about 1kHz, typically). Below about 500 Hz bass can be narrow-band equalized - attenuating peaks, but ignoring narrow dips.
It is amusing that the ancient B&K target curve is still trotted out. Read the original paper and see the frequency responses of the grossly colored speakers used in their test - typical of the time (1972 I think), but nowadays nobody in this forum would give them house room - loudspeaker technology has advanced. That said, as a broad "average" curve, the tendency is correct.

The room curves associated with the best sounding conventional forward-firing loudspeakers are all downward tilted - why? Remember the curve is "an end result, not a target" and conventional loudspeakers are omnidirectional at very low frequencies, becoming more directional as frequency rises. So, if such loudspeakers are designed with desirable flat on-axis/listening window frequency responses (the first arriving "direct" sound) the total sound output decays with increasing frequency. Steady state room curves measure direct + all reflected sound so they must tilt downward - or, more accurately, rise towards the bass. Then there is air absorption, which adds more downward tilt above about 1 kHz. It is all understood and predictable. Also understood is that the detailed shape of a room curve depends on which loudspeaker is used. Even if designed to be flat on axis the off-axis performance is the dominant factor in determining the shape of the curve - equalization cannot, can not, compensate for uneven off-axis sound radiation. Only a better loudspeaker can do that. If you search my responses in this forum in recent years you will find more explanation. That is why we should be grateful to see useful technical measurements (called the spinorama) on this and other forums. Starting with a well designed loudspeaker will yield a smoothish, slightly tilted room curve at the listening position - except at low frequencies and that is a separate story, with multiple solutions. It is not a target curve, though, and adjusting flawed loudspeakers to imitate the shape will not yield equally good sound. Sorry.

Finally, there are the inevitable variations in spectral balance in recordings. This is why one needs tone controls, if one is fussy.

So, if you have found "perfection" in sound, good for you. Enjoy it. But, your room curve tells me nothing useful unless I can see a spinorama or equivalent anechoic data on the loudspeaker that produced it. Otherwise the forum discussions of what curve is right or wrong cannot end.

How would a person use the spinorama data to then use EQ to produce a desired house curve for the given speakers? Does that mean a person should not use EQ to change a slight v shaped response curve into a slightly down-tilted sound curve that loses say 10dB from 20Hz to 10KHz?

Sorry if those are obtuse questions or if I'm missing something.
Thanks.

 

[ernestcarl]

How would a person use the spinorama data to then use EQ to produce a desired house curve for the given speakers? Does that mean a person should not use EQ to change a slight v shaped response curve into a slightly down-tilted sound curve that loses say 10dB from 20Hz to 10KHz?

Sorry if those are obtuse questions or if I'm missing something.
Thanks.

At the end of many (not all) speaker reviews Amir or Maiky gives suggested PEQ filters that one can test out. I would not always follow those for my own personal use case, but it’s a start.

For example: https://www.audiosciencereview.com/...s-rainmaker-speaker-review.40906/post-1444412

I would create prediction models out of the provided data (attached at the end of the review) and compare it to my own objective measurements and subjective listening results. I’ve done something of the sort for the Fluid FX50 before not so long ago.

Post your EQ settings and state your transducer too.

I am using EQ APO with Sennheiser HD 598SR headphones.

www.audiosciencereview.com

 

[LTig]

How would a person use the spinorama data to then use EQ to produce a desired house curve for the given speakers? Does that mean a person should not use EQ to change a slight v shaped response curve into a slightly down-tilted sound curve that loses say 10dB from 20Hz to 10KHz?

Sorry if those are obtuse questions or if I'm missing something.
Thanks.

You don't use EQ to get a house curve. Use the spinorama to select a speaker with smooth dispersion (without such you're lost) and optimally a flat and smooth frequency response. If it's not flat you can EQ it like @amirm does in hiS reviews.Then measure frequency response and flatten the peaks below the Schroedinger frequency (about 500 Hz) to fix room problems (room EQ).

This done you'll end up with a frequency response which - depending on the reflection character of your room and the off axis frequency response of your speakers at all angles - may look more or less close to a house curve. If it's not to taste better treat the room than use more EQ.

 

[oversky]

Put differently, an omni microphone is not equivalent to two ears and a brain.

Will In-ear BINAURAL MICS or headphone fixture such as GRAS 45CA, better than an omni microphone in the room measurement?