Possible to rolloff treble with Dirac?

[rcohen]

I have some new speakers with an extended treble response, so there is no natural rolloff at 20khz.

 

They are sounding harsh to me, so I'd like to EQ down the top end.

 

Unfortunately, I have no way to do this in Dirac, unlike my past speakers that had a natural rolloff at 20khz.

 

That's because Dirac blends out the target curve on the right side of the graph. So, if I try to make a target curve that slopes down, the graph shoots back up at 20khz.

 

Any suggestions? Is there a way to make Dirac act like a shelf filter to the right of the target curve, rather than just passing the sound unaltered?

[RFP]

I have some new speakers with an extended treble response, so there is no natural rolloff at 20khz.

 

They are sounding harsh to me, so I'd like to EQ down the top end.

 

Unfortunately, I have no way to do this in Dirac, unlike my past speakers that had a natural rolloff at 20khz.

 

That's because Dirac blends out the target curve on the right side of the graph. So, if I try to make a target curve that slopes down, the graph shoots back up at 20khz.

 

Any suggestions? Is there a way to make Dirac act like a shelf filter to the right of the target curve, rather than just passing the sound unaltered?

 

I would suggest that the response above 20KHz is not a cause of the audible harshness. Indeed, a truly great wide-band system should sound as smooth as silk. The likely cause is some excess energy lower down in the treble range, or failing that, the new speakers might be revealing some shortcomings upstream.

 

In any case, if you really must roll off the response at 20kHz, there are more appropriate ways of doing it than using Dirac Live.

 

The easiest method is to listen to the speakers slightly off-axis.

[rcohen]

I would suggest that the response above 20KHz is not a cause of the audible harshness. Indeed, a truly great wide-band system should sound as smooth as silk. The likely cause is some excess energy lower down in the treble range, or failing that, the new speakers might be revealing some shortcomings upstream.

 

In any case, if you really must roll off the response at 20kHz, there are more appropriate ways of doing it than using Dirac Live.

 

The easiest method is to listen to the speakers slightly off-axis.

I'm referring to 20khz and down, not above 20khz.

Why is EQ software not an appropriate way to tune the EQ to my preferences?

In the past, I've had great results fine tuning the top octave in Dirac.

[SteveS1]

Can you post your graph?

[wgscott]

Why can't you use a customized target curve. The more points you add to it, the more control you have over slope and inflection point, etc, and you also should be able to move the boundaries. Remember, unless you are a young child, you aren't hearing much above 16kHz anyway...

[dallasjustice]

I have some new speakers with an extended treble response, so there is no natural rolloff at 20khz.

 

They are sounding harsh to me, so I'd like to EQ down the top end.

 

Unfortunately, I have no way to do this in Dirac, unlike my past speakers that had a natural rolloff at 20khz.

 

That's because Dirac blends out the target curve on the right side of the graph. So, if I try to make a target curve that slopes down, the graph shoots back up at 20khz.

 

Any suggestions? Is there a way to make Dirac act like a shelf filter to the right of the target curve, rather than just passing the sound unaltered?

 

Can you drag the limit of the HF correction out to a higher frequency? Maybe you inadvertently had it dragged down in frequency and you need to move it back up again? Just a guess on my part.

[rcohen]

Thanks for helping. Finally got a chance to follow up.

 

Here are some graphs to illustrate the issue better.

 

 

My old speakers roll off before 20khz, so Dirac allows me to draw a target curve with the downward slope that I like.

 

My new speakers have an extended treble response, so I can't slope the target curve down while keeping it lined up on the right side. This sounds too bright for me in my room. Never mind the dip at 200hz. That's caused by bad room placement.

 

Here's what happens if I try to slope down the target curve. I can't find any way to drag the right edge down to make a shelf filter. I wouldn't really want to slope it down this much, but you get the idea.

[flak]

Hi Robert

 

it seems to me that in this case we are more into speaker voicing than room acoustics... if I understood correctly the desired frequency response at the high frequencies is very different from the actual speakers response as measured by Dirac Live.

I think that you should not create a steep roll-off but actually drag down the higher curtain to a point where there is not much change in the filters (where the target is rather flat and aligned with the measurement) to avoid any steep changes in the filter in that region.

 

Apparently the previous speakers did not have that large drop around the 200 Hz area vs. the other regions of frequencies, DL is trying to align that to the rest of the frequency response while you would like to align all of the higher frequency response to that dip.

Dirac Live will not create filters that affect the area above 20kHz so probably you'd better use adjustments in the speakers themselves to roll off the highs if you have them, or may be you could use some kind of EQ.

 

my two cents,

Flavio

 

p.s. which speakers have replaced your previous ones?

[rcohen]

I kind of have a revolving door for used speakers. The grass is always greener.

 

The old set are Triad Platinums. The new set are JTR 212HTs.

 

I'm pretty sure that the 200 Hz dip on the JTRs are caused by cancellation off the walls. I have them temporarily positioned about 2.5 feet from both the rear and side walls. I need to try moving them and/or moving some acoustic panels.

 

I am more concerned with recreating the overall FR tilt that I had before than the exact shape of the FR. Perhaps I will be able do that by fixing that 200 hz null and cranking up my subs by 5-10 db.

 

It would be nice if DL would allow me the option to drag that last control points up and down and treat them as a shelf filters, rather than forcing no filter. That way, I wouldn't lose the ability to do this type of EQ on speakers with extended treble, since I agree that the steep filters seem like a bad idea.

[rcohen]

I'm still having this problem.

 

For my new speakers that don't have a big rolloff at 20 kHz, Dirac won't let me tune the curve for my desired frequency response tilt.

[Danold]

I'm still having this problem.

 

For my new speakers that don't have a big rolloff at 20 kHz, Dirac won't let me tune the curve for my desired frequency response tilt.

 

Why don't you tune the front end (lower frequencies, e.g. 20-500 Hz) up instead of the higher freqs down? As Dirac has a playback attenuation of 8db across the whole band there shouldn't be a problem .

[rcohen]

Why don't you tune the front end (lower frequencies, e.g. 20-500 Hz) up instead of the higher freqs down? As Dirac has a playback attenuation of 8db across the whole band there shouldn't be a problem .

Then I end up with a steep filter on the low end, and associated ringing.

 

All I can figure out is to crank up the volume on the subs so that I can avoid steep slopes on both sides. The problem is that I have to remeasure every time I iterate on the right sub level, which is very time consuming. Again, that could be avoided with the ability to do a shelf filter on the low end.

[rcohen]

I just realized that Dirac will let me zoom in to the right side of the graph, past what you can see in the 1x view. It lets me drag the limit on the right side of the graph out further, past the point where my mic rolls off. This gives me more flexibility when drawing the curve. Does this cause problems? It seems like it will help, but I'm worried since I might be exploiting a bug.

 

[rcohen]

BTW, cranking up the sub didn't work, since it left a big dip in the middle, and Dirac seems to limit boost to 10db. I need a method that lets me get the right side down.

 

Another approach I could use is to apply EQ either in JRiver or in my preamp.

 

Interestlingly, the new Dirac seems to let me use JRiver ASIO as the playback device during calibration. If I were do do EQ and/or bass management in JRiver, rather than my preamp, is that the preferred method?

 

It comes back to the old question of whether it's better to do EQ and bass management before Dirac or after Dirac.

[rcohen]

One concern about calibrating Dirac through JRiver is that the order reverses.

Dirac -> JRiver for calibration

then

JRiver -> Dirac for playback

[beanbag]

It comes back to the old question of whether it's better to do EQ and bass management before Dirac or after Dirac.

 

Do the management BEFORE Dirac, so that Dirac can "fix" any phase problems you created.

[rcohen]

Do the management BEFORE Dirac, so that Dirac can "fix" any phase problems you created.

Wouldn't it have to happen after Dirac, so that Dirac can measure it?

[beanbag]

Dirac isn't on when you are measuring.

[rcohen]

Dirac isn't on when you are measuring.

 

By "it" I meant JRiver.

[rcohen]

Following up on this thread, the approach of zooming in on the right side of the graph and extending the EQ window as shown in post #13 gave me great sounding results that I was very happy with.