What is your favorite house curve

[TimVG]

My favourite house curve involves picking a loudspeaker with a good amplitude response and directivity characteristics, and doing absolutely nothing* above a couple hundred of Hz and simply adjusting the response as to follow the natural slope down below the transition frequency.



Steady state measurements can portray (or hide depending on the neutrality of the direct sound) directivity issues, and other artifacts such as crossover dips caused by vertical lobing. High frequency spectral balance will depend on the speaker and its characteristics, room (size), listening distance.

The reason a certain target curve was preferred in the Olive tests was because the speaker tested was not neutral to begin with, and also because bass accounts for 30% of preference ratings. Had they flattened that B&W from an anechoic POV, and give it a similar 'low end heft' - things may have turned out differently.


*I did produce some semi-anechoic measurements to flatten out the listening window above 1kHz

 

[JoachimStrobel]

Is it then correct to assume that:
Room curves are needed to recreate a music hall environment at home.
In a music hall, frequencies above 2khz emitted from an instrument see a reduction in amplitude due to dispersion and reflection before they reach the listener.
A room curves recreates this high frequency fall-off for home listening.
It is not clear to me, if the off-axes response from loudspeakers are just coincidentally causing the same roll-off.
Seems that small jazz venues with acoustic instruments might then sound bright, whereas classical music in large halls may have a stronger fall-off.

But what about the low frequency? Does a bass loose amplitude at a typical listening position in a concert hall?
Toole‘s room curve shows a healthy bass boost. Or is that a compensation for a loss in bass amplitude between a 100db concert strength and a 60 dB home listening level, hence the difference between the Iso226 curves for the said levels?
That would be 20db or so for low frequencies while Toole‘s curves is around 6-8db, others are flat. So what is the theory behind that?
Likewise, there is a 10 dB or so difference for high frequencies between the 100db and 60db loudness curves. Is this build into Toole’s curve, hence it is dipping a bit less as needed for a sole dispersion&reflection correction. Is that room curve then specific for a certain listening different level?
Comments?

 

[TimVG]

I remember @Floyd Toole mentioning his own steady-state curve is 'flattish' in the bass range - not sure where you got the 'healthy bass boost' idea from. In any case - a room curve/house curve .. is not a target in the sense that if you don't have it, you can EQ your speakers it in-situ to match it. It simply doesn't work like that. It may make a non-neutral loudspeaker better, but the only sure thing are anechoic measurements.
Good loudspeakers will display decent room curves above the transition frequency, but high-frequency drop-off will depend on your listening distance, general dispersion of the loudspeaker in question and the amount of absorbing material present.
My own curve above is uncorrected above a couple hundred Hz - Below that I tried to made it follow what the speakers were already, naturally, doing. 4 subs help dig down deep. I don't feel the need for any additional boosting

 

[JoachimStrobel]

Good loudspeakers will display decent room curves above the transition frequency.

I agree that they do that - but the mechanism is not clear to me.
A trumpet played in a concert hall will have its high frequency rolled off due to dispersion&reflection. And why does the well-behaved off axes response of a good loudspeaker do the same? Again, just coincidence?
Not correcting above 300-500hz works good for Stereo, but for multichannel, achieving tonal balance may be a reason to correct for higher frequencies too.

 

[Floyd Toole]

You need to look at this thread for much more information about room curves:

• Forums
• Audio, Audio, Audio!
• Psychoacoustics: Science of How We Hear

An Enticing Marketing Story, Theory Without Measurement?

 

[TimVG]

Good loudspeakers will display decent room curves above the transition frequency.

I agree that they do that - but the mechanism is not clear to me.
A trumpet played in a concert hall will have its high frequency rolled off due to dispersion&reflection. And why does the well-behaved off axes response of a good loudspeaker do the same? Again, just coincidence?
Not correcting above 300-500hz works good for Stereo, but for multichannel, achieving tonal balance may be a reason to correct for higher frequencies too.

One has nothing to do with the other - dispersion patterns of instruments can be quite complex. I play the horn, the sound an audience, that is seated some distance away, hears, is entirely reflected. Accurately portraying the timbre of the instruments, capturing the 'sound' of the hall (ambiance), those are the result of things that occur prior to playback: microphone type and placement, mixing, mastering, .. if the room were crucial to the timbre, headphones would be a mess - which they aren't.

The reason tonal balance can be off, you mention multichannel, is because partially because of certain standards, wrongfully implemented sometimes. The X-curve is an example. But there is also the simple lack of objective standards in the recording world. Thus, recordings vary - a lot sometimes.

 

[JoachimStrobel]

I remember @Floyd Toole mentioning his own steady-state curve is 'flattish' in the bass range - not sure where you got the 'healthy bass boost' idea from.

That is the one from Toole´s open source PDF.

 

[Thomas_A]

View attachment 34506
That is the one from Toole´s open source PDF.

So a conclusion is that even if you have the ideal loudspeaker in the ideal normal room producing the idealised steady state room curve, neither the trained nor the untrained listener would like the result (?!)

 

[Krunok]

A room curves recreates this high frequency fall-off for home listening.

Loudspeaker designers tilt response upward just a little in anechoic environment so you get in-room response tilted down just about right when measured on-axis at your LP. That strategy is based on research what kind of curve people preffered most. The thing that is equally important is however to have smooth of-axis response as well, but unfortunately you can only hope they did that well as usually no data is provided.

As an example, take a look at F208 anechoic response (on-axis is black topmost line):

As an in-room example, here is response of my Harlechs measured from 15cm on axis (upper curves) vs LP (lower curves):

 

[TimVG]

As said before, people prefer anechoically neutral loudspeakers in blind tests - the measured steady-state curve is the result of loudspeaker / room / listening location in said room. Above the transition frequency, it can be accurately predicted. So, two statistically tied neutral loudspeakers, can still produce different room-curves - so is one more correct then the other? Of course not. In terms of spectral balance, the recording can be decisive - but we can't adjust our entire system for every recording right? That's where bass/treble settings come in handy.

 

[Krunok]

As said before, people prefer anechoically neutral loudspeakers in blind tests - the measured steady-state curve is the result of loudspeaker / room / listening location in said room. Above the transition frequency, it can be accurately predicted. So, two statistically tied neutral loudspeakers, can still produce different room-curves.

I don't really think so. If you have 2 loudspeakers with very similar spinorama charts and you put them in the same room at he same location I believe they will have very similar in-room response. I also believe their SQ will be perceived very similar by most people. That is the point of Toole's research.

 

[TimVG]

I don't really think so. If you have 2 loudspeakers with very similar spinorama charts and you put them in the same room at he same location I believe they will have very similar in-room response. I also believe their SQ will be perceived very similar by most people. That is the point of Toole's research.

Speakers can still be neutral on and off axis, yet feature different directivity indices (let's say very directional horn system vs cone/dome system) this can be evident in the steady state curve

 

[Krunok]

Speakers can still be neutral on and off axis, yet feature different directivity indices (let's say very directional horn system vs cone/dome system) this can be evident in the steady state curve

How exactly would very different directivity index be evident in the steady state (I pressume in-room) curve?

 

[TimVG]

How exactly would very different directivity index be evident in the steady state (I pressume in-room) curve?

Since the predicted room curve is calculated from the direct sound, early reflections and sound power curves - it is logical that different directivity will produce different curves.

 

[LTig]

How exactly would very different directivity index be evident in the steady state (I pressume in-room) curve?

Wouldn't the angle of the falling steady state curve be steeper when the speaker has higher directivity (less highs off axis)?

 

[Krunok]

Wouldn't the angle of the falling steady state curve be steeper when the speaker has higher directivity (less highs off axis)?

It would, but in his example with horn tweeter and woofer vs dome tweeter and woofer you wouldn't really have similarly smooth directivity index at the XO region as woofers would behave similar while tweeters wouldn't.

 

[TimVG]

It would, but in his example with horn tweeter and woofer vs dome tweeter and woofer you wouldn't really have similarly smooth directivity index at the XO region as woofers would behave similar while tweeters wouldn't.

I should have been more clear. I intended to refer to a 3 or 4-way cone/dome system. A CD horn/compression driver is usually coupled to a larger woofer (to provide constant, but more limited dispersion) as opposed to a cone/dome system where one would couple the tweeter to a smaller mid to maintain wider directivity. This can be evident when comparing these systems in a room and taking steady-state measurements. Another thing that will tend to show in the latter system is a directivity dip in the higher treble due to vertical cancellation effects between tweeter and mid.

 

[JoachimStrobel]

My take is: Dispersion and early reflection, not ambience, causes a instrument specific roll off at high frequencies. Instruments have smooth off axes (wherever that is) response, no dents in the frequency pattern, hence the roll-off may be different for different instruments, but it is smooth ( Ambience comes on top, but open air venues and small jazz clubs do not have it)
Good loudspeakers have a smooth off-axes response and hence cause a smooth roll-off. Bad speakers have dents in their off-axes response, and after convolution with dispersions and reflections that dent has migrated and can hence not be Eqed back.
However, the magnitude of the roll-off differs from room to room. Therefore, a good loudspeaker might need some help of a DSP to adjust it - to what?. One would need a RTA readout of an instrument playing in the living room to check the roll-off needed. I am surprised that such data does not exist.
I have 6 loudspeakers of the same type for multichannel. But my room is a living room and hence they stand where there is room for them. Hence they need EQ for high frequency. There is no way around. And there is the projector’s screen covering the centre speaker(s) too. Hence a good high frequency room curve is important. I listen to Sacd, DVD-Audio and DTS 5.1 squeezed on a CD. I hope that no X-curve is used. I am sure it is not.
True 5.0 mixes have a lot of direct sound. The HTF gets important too. In my setup, the rear speakers shine directly onto my eardrums - something an Umik does not measure...[/QUOTE]

 

[JoachimStrobel]

Loudspeaker designers tilt response upward just a little in anechoic environment so you get in-room response tilted down just about right when measured on-axis at your LP. That strategy is based on research what kind of curve people preffered most. The thing that is equally important is however to have smooth of-axis response as well, but unfortunately you can only hope they did that well as usually no data is provided.

As an example, take a look at F208 anechoic response (on-axis is black topmost line):



As an in-room example, here is response of my Harlechs measured from 15cm on axis (upper curves) vs LP (lower curves):

View attachment 34524

I get that and for me for the high (>1 kHz) part I am done with. The low frequency part is interesting. What are we correcting for? Certainly not dispersion. And opinions do differ, from 0 to +12dB, and that is after room modes are mostly taken care of. In my opinion some corrections are a kind of loudness correction, correcting for differences between a 120 dB concert and a 60db listening room. But I really do not know, but I agree with Toole: Bass defines how we like the sound. ( and I did read most of the suggested posts and his book and more....)

 

[Thomas_A]

My take is as usual that it is that the speaker response sets the base. A flat or near flat on axis with smooth off-axis (as I prefer a bit less energy in the 2-5 kHz region due tot the stereo system errors, cf also the F208 on axis curve above). The reflections from the listening venue should not be too few or too early, and smaller rooms where the side walls are to close may need more narrow dispersion and/or more toe in to reduce the level of reflections in the 2-10 ms window. Reflections from behind the speakers should be avoided (so I damp these) since there is no benefit in colouring the direct sound from the recording (i.e. at 0 degrees). I only EQ the room modes below 100-150 Hz or so. Below is my room result (raw) 200 Hz-22 kHz, with the "Toole idealised" curve superimposed.