Some comments from Floyd Toole about room curve targets, room EQ and more

[Robbo99999]

Toole says not to EQ a crossover dip if it disappears under angles and thus at the sound power, so usually at loudspeakers with not smooth directivity. Do you want to EQ your JBL 308P MKII? If you want you can post your measurements, but I strongly doubt the 308 has a significant problem at its crossover frequency that should be corrected, especially since the MKI didn't have one.

(For the moderator, I hope these posts are not off topic, I don't believe them to be as I'm attempting to EQ my system based on the recommendations/guidelines established by Toole, which is the topic of this thread, and from the points learned in this thread.)

I've done some measurements creating average graphs of 3 positions measured within 20cm of the ideal listening position, I've use Var Smoothing option in REW. I've done the above for both stock speakers with no Low Shelf Boost, then one for 2dB Low Shelf Boost at 110Hz (Q=1), then another one for 3dB and 5dB also. These are also with time delays to match the 2 speakers (left speaker 20cm closer), and also matched channels to equal dB so created a -0.4dB offset for the closer left channel. What are your thoughts on which curve to choose (in terms of Low Shelf Boost) as a starting point for EQ, and which peaks (and/or troughs) would you EQ out, and what do you think we can learn from these graphs...and is it even sensible to use Low Shelf boost and then EQ out the created peaks? These are all based on my "Half Harman Curve" that I would use for TV/Movie viewing.

Stock Speakers (no Low Shelf Boost):

2dB Low Shelf Boost 110Hz Q1:

3dB Low Shelf Boost 110Hz Q1:

5dB Low Shelf Boost 110Hz Q1:

 

[QMuse]

(For the moderator, I hope these posts are not off topic, I don't believe them to be as I'm attempting to EQ my system based on the recommendations/guidelines established by Toole, which is the topic of this thread, and from the points learned in this thread.)

I've done some measurements creating average graphs of 3 positions measured within 20cm of the ideal listening position, I've use Var Smoothing option in REW. I've done the above for both stock speakers with no Low Shelf Boost, then one for 2dB Low Shelf Boost at 110Hz (Q=1), then another one for 3dB and 5dB also. These are also with time delays to match the 2 speakers (left speaker 20cm closer), and also matched channels to equal dB so created a -0.4dB offset for the closer left channel. What are your thoughts on which curve to choose (in terms of Low Shelf Boost) as a starting point for EQ, and which peaks (and/or troughs) would you EQ out, and what do you think we can learn from these graphs...and is it even sensible to use Low Shelf boost and then EQ out the created peaks? These are all based on my "Half Harman Curve" that I would use for TV/Movie viewing.

Stock Speakers (no Low Shelf Boost):
View attachment 59427
2dB Low Shelf Boost 110Hz Q1:
View attachment 59428
3dB Low Shelf Boost 110Hz Q1:
View attachment 59429
5dB Low Shelf Boost 110Hz Q1:
View attachment 59430

I'm not OP but will offer my opinion: last one defintiely looks the best to me. I draw 10dB falling line for easier comparison.

 

[Robbo99999]

(note I'm aiming for "Half Harman Curve" which is not as severe as your red line trend there)

(I will aim for "Full Harman Curve" for music though, but my first port of call is Half Harman Curve for TV & Movie which is my priority).

 

[QMuse]

(note I'm aiming for "Half Harman Curve" which is not as severe as your red line trend there)

(I will aim for "Full Harman Curve" for music though, but my first port of call is Half Harman Curve for TV & Movie which is my priority).

Ah ok, I see. Still the last one looks the best to me.

 

[thewas]

Bass level strongly depends on personal preferences, listeniung level, room modes and material, so if I was you I would chose the one that makes most of the music you listen to sound enjoyable to you.

 

[Robbo99999]

Bass level strongly depends on personal preferences, listeniung level, room modes and material, so if I was you I would chose the one that makes most of the music you listen to sound enjoyable to you.

Note, those curves I posted are before EQ bar the Low Shelf boost I mentioned. If you remember I was talking initially about concerns re EQ'ing the area in the crossover region (1800Hz), but I also wanted to get some pointers from you re "best practice" in EQ'ing with regards to the specifics of my curves. Yeah, so those graphs you see, I've not applied the auto EQ yet, I plan to remove the peaks. At the moment the 2dB Low Shelf graph looks like the best starting point to EQ out all the peaks in the 20-550Hz region. But I was also wondering if it was good or bad practice to apply a Low Shelf Boost and then EQ out the peaks (double equalisation if you will)? So I'm looking for some general pointers / specific pointers in relation to the graphs I posted. Note they're all based around the target of a "Half Harman Curve", which I've coined to be half the bass boost & half the treble loss of a normal Harman Curve...from personal experience I know that Half Harman Curve sounds best to me in TV/Movies, so I'm looking for specific EQ advice rather than choice of target curve (the blue line is obviously the target which is the same in all the graphs).

EDIT: and to me it looks like the sharp dips at 75Hz, 110Hz, and 130Hz should be left alone as being unEQ'able? Do you think the peak at 1800Hz, which is also the crossover point, do you think that point should be EQ'd down flat, or best left alone considering it's at the crossover point and also we don't know if that's based on anechoic or room influence, with thinking being you should really only EQ "anechoic evidence" above the transition point?

 

[thewas]

From gain and EQ depth its usually good practise to use a the shelf settings which give you the closest curve to your desired target so you will need only minimal additional (P)EQ gains.

 

[Robbo99999]

Ah ok, I see. Still the last one looks the best to me.

Is that because some of the troughs have been brought up closer to the target?

 

[QMuse]

(note I'm aiming for "Half Harman Curve" which is not as severe as your red line trend there)

I corrected my red target line to "Half Harman Curve" - last one still looks like a champ!

 

[QMuse]

Is that because some of the troughs have been brought up closer to the target?

Yes. EQ-ing that curve to your target should be least obtrussive.

 

[Robbo99999]

From gain and EQ depth its usually good practise to use a the shelf settings which give you the closest curve to your desired target so you will need only minimal additional (P)EQ gains.

Cool, I kind of intuitively thought a Low Shelf Boost made sense when I was having to deal with only a limited number of available filters in miniDSP - 12 filters. But it sounds like it's also good practice in terms of minimising the potential negative sonic effects of numerous peak filters.

I edited my post you quoted after you quoted my post, can you have a read of my EDIT points and see what you think?

 

[Robbo99999]

I corrected my red target line to "Half Harman Curve" - last one still looks like a champ!

View attachment 59443

Yes. EQ-ing that curve to your target should be least obtrussive.

Ah, in that case wouldn't it make sense to lower that red line of yours so it sits in the area of my blue line, that way you can EQ out the peaks and have less troughs?

 

[QMuse]

Ah, in that case wouldn't it make sense to lower that red line of yours so it sits in the area of my blue line, that way you can EQ out the peaks and have less troughs?

Absolutely, I drove it in "envelope" style just for comparison. When EQ-ing you would want it somewhere here:

 

[thewas]

Ah, in that case wouldn't it make sense to lower that red line of yours so it sits in the area of my blue line, that way you can EQ out the peaks and have less troughs?

Mind you that in the modal region we hear more peaks than dips so the target curve should be closer to the peaks than in the middle of peaks and dips. A good way to test if the overall tonal balance is close to your target curve is the psychoacoustic smoothing of REW (but not to create EQs).

 

[Robbo99999]

Here's the before & after, I've finished EQ'ing for this "session" today, I chose the 2dB Low Shelf Curve from one of my previous posts as I ran out of available filters to EQ the 3dB and 5dB Low Shelf Curves, and it allowed more headroom for volume, less overall boost....EQ'ing only the 20-550Hz region:

Before EQ:

After EQ (actual measurement) - 3dB of overall Input Gain Lost:

And I also have this following graph as a option too, which is the same as the above except with allowing REW to boost the little trough at 60Hz, but that cost an extra 2dB of Input Gain Loss:
After EQ (actual measurement) - 5dB of overall Input Gain Lost:

Subjective Listening Experience:
In terms of listening to scenes from the film Prometheus back to back comparing those 2 curves, it's a bit of a toss up between them. There is a feeling of slightly fuller bass from the last graph, but there is slightly more detail to be heard in the 2nd graph...these were played back at same volume level (I chose the same Input Gain Loss in both for this listening experiment).

Conclusions & Questions:

• I'm leaning towards using the 2nd graph above more than the 3rd, because it's 2dB less Input Gain loss and there is slightly more detail to be heard in general using that one (for some reason), with the trade-off being it's slightly less full in the bass.
• Using Var Smoothing is very very unforgiving on the bass area (thereby showing every tiny sharp dip & peak), is there any merit in using more smoothing in that area to potentially save on number of filters used?
• I think I've chosen the right peaks and area to EQ right? The 20-550Hz area?
• It looks to me like the dip at 75Hz / 110Hz / 135Hz are all impossible and right not to EQ?
• I also couldn't get rid of the peak at 120Hz, I think it's because it's slap bang inbetween two sharp dips.
• With miniDSP I had 12 filters at my disposal, and I used them all so I have nothing left over to EQ above 550Hz, do you think it's wise not to EQ above that area from what you see in my graphs anyway?
• When I use Equaliser APO for music listening (the previous is all talk re TV/Movie watching), I'll have more filters available, so it's possible I could EQ the curve above 550Hz, do you think this would be wise though? The broad peak at 1800Hz looks like it could be brought down, but that's right at the crossover point. Also, in light of the lack of anechoic data for this JBL 308p Mkii, is it wise to leave this area unEQ'd too?

Yeah, so I've got a few questions there, I would be grateful for people's thoughts & analysis on it.

 

[thewas]

Looking good, if you want to have less filters it's often better to create them manually as the REW auto EQ while clever doesn't have our intelligence. In the beginning it takes some time but after some trials it gets faster, nowadays I only need 10 minutes for a set of filters.

About the narrow modal dips, it makes sense not to fill them like you do

About the mid peak, just create 2 eq versions with and without EQing it and compare it with some good recordings you know, especially with human voices.

 

[Robbo99999]

Thanks, I could try experimenting with some manual filters. In fact I loaded REW up just now whilst typing this post and it looks like you can create manual filters in REW and see the effect as you change them, so that's fairly easy. I'll leave that for another EQ'ing session to see if it means I can save a filter for EQ'ing out the 1800Hz peak....which I will then compare back to back during listening tests to see if there is an improvement or degredation with EQ'ing that 1800Hz area. Previously I have EQ'd the entire frequency curve when using Equaliser APO (unlimited number of filters effectively), and I've been quite happy with it, but given what I've read here (& elsewhere) there are lots of calls for brevity when it comes to EQ'ing above the transition frequency....so I wanted to experiment with that today by just EQ'ing the 20-550Hz area and also to check with you guys that I was taking the right approach. It sounds like I am.

 

[Robbo99999]

@QMuse , you said you thought the 5dB Low Shelf curve was the best, but I had the problem that I didn't have enough miniDSP filters available to EQ that curve....well I spent 10 mins just now tweaking the 5dB curve and managed to eject one of the automated REW filters (it wasn't really necessary), so that allowed me to use that filter for 5dB Low Shelf. It is actually better for movies given I don't have a sub - the 5dB Low Shelf provides enough extension into the low bass (below 40Hz) that enables a bit more rumbling atmospheric impact to movie sound tracks. Apart from filling out the low end more below 40Hz it's not really any different (looks the same) to the 2dB Low Shelf Curve that I had decided on a few hours ago (because it's to the same Half Harman Target Curve). (Costs me 5dB Input Gain though, which is 2dB more than the one I decided on a few hours ago, but I think it's worth it).

EDIT: it also seems to balance the overall tonality more, which I'm surprised about as it mainly seems to be extending the response below 40Hz, and perhaps filling in some of the troughs just a little more above that point - combined seems to have pushed it past the tipping point into significantly better sound.

 

[Robbo99999]

Just a fairly quick question about room curves....different real rooms would have different natural room curves wouldn't they, in terms of how tilted they would be if an anechoically flat speaker was playing in it? The Harman Curve is based on their listening room right? Who's to say you would only prefer one type of room for music, maybe different people prefer the feels of different rooms, or different rooms might suit different styles of music better. Maybe if it was possible to determine your actual real room curve for whatever listening room you find yourself in - perhaps mimicking that would be better? I suppose you could find out what your real room curve was by EQ'ing flat a speaker from Amir's measurements and then measuring that speaker in your room from the listening position and noting what the tilt of the curve was.....that would be your own room's room curve right?

EDIT: or thinking about it, perhaps your room curve target should be whatever target was used by the recording engineers of that artists music....if you wanted to reproduce it faithfully? But then how would you know what that would be....I suppose you'd take a sensible approximation of that somehow if the info was out there for measured recording studios?

 

[thewas]

Just a fairly quick question about room curves....different real rooms would have different natural room curves wouldn't they, in terms of how tilted they would be if an anechoically flat speaker was playing in it? The Harman Curve is based on their listening room right?

To both questions the answer is yes.

I suppose you could find out what your real room curve was by EQ'ing flat a speaker from Amir's measurements and then measuring that speaker in your room from the listening position and noting what the tilt of the curve was.....that would be your own room's room curve right?

But only for the specific directivity of that loudspeaker as the generated response at the listeners position also depends on it.

EDIT: or thinking about it, perhaps your room curve target should be whatever target was used by the recording engineers of that artists music....if you wanted to reproduce it faithfully?

Unfortunately http://seanolive.blogspot.com/2009/10/audios-circle-of-confusion.html and the reason why many older recordings rather tend to sound bass shy (being mixed with not room-corrected large monitors) but if that circle is overcome and all recording engineers just use neutral monitors corrected in the bass region to their room than we won't have that variable any longer.