TOPIC:

Hello,

I am experiencing some issues in my room with regards to the phantom center channel in my stereo setup. Speakers are positioned symmetrical in a non-symmetrical room (left side has a dormer, right side behind the listering position has a more open space). Without any correction from my side (changing settings in SHD or moving the left speaker closer) the center is always off-center and shifts to the right. I noticed that putting a time delay of 0.44 ms for the right channel resolves my issue and shifts the center of the soundstage as I'd like. I also tried to move the left speaker closer to the listening position, but this results in weird unbalanced soundstages (and dirac compensates as well by lowering the gain of the closer speaker, resulting again in issues).

As the SHD manual suggest this should be done before using dirac live. 
I quote 8.3.7 Time delay:  Note: The Dirac Live analysis algorithm also sets a time delay on the left and right channel. The time delay on the output channels should be used to time-align drivers for a two-way loudspeaker or to optimize subwoofer integration before doing Dirac Live calibration.

However after using dirac live, my output settings are overruled and the center of the soundstage shifts again off-center to the right. Any option to set output time delays which are used as basis and not becoming overruled by dirac? 

Many thanks

There is an option in Dirac Live to disable its time delay adjustment, so if you prefer what you have set manually then you can use this to keep things as they were.

For info. what  I believe Dirac Live does is to use the arrival time from each speaker at the central measurement position to set the time delay, so if you've deliberately set a delay to achieve something else then it is totally to be expected that Dirac would cancel this out. 

hannesb wrote:

As the SHD manual suggest this should be done before using dirac live. 
I quote 8.3.7 Time delay:  Note: The Dirac Live analysis algorithm also sets a time delay on the left and right channel. The time delay on the output channels should be used to time-align drivers for a two-way loudspeaker or to optimize subwoofer integration before doing Dirac Live calibration.

Just to clarify this point, the point being made is that Dirac Live can only adjust the left vs right delay, but cannot adjust the delay between subwoofer and main speakers, or low and high-frequency drivers in a single speaker. 

Thank you for your answer. I found an option to disable gain/delay in dirac and saw indeed a difference regarding the impulse respone tab. When the filter is exported to the SHD, it looks like no gain/delay is set on the input/dirac side of the SHD, however, somehow it looks like it overwrites the initial defined delay on the output side back to 0. If nothing is changed after exporting the filter, the center shifts back to the right, even though delay is still having the same settings in the SHD. Once you adjust and set it back to the initial settings to correct, you hear the center shifting towards the center, only the frequency response graph looks afwul in REW afterwards...

It looks like there is a bug or so. 

hannesb wrote:

 only the frequency response graph looks afwul in REW afterwards...

It looks like there is a bug or so. 

Almost certainly not a bug. What you need to bear in mind is that Dirac Live deliberately targets an average response across the measurement volume, not the response at any single point where you place your microphone to make a measurement in REW. As a test, it is actually possible to make DL produce filters by only making one measurement at the central location. If you do this and then make a measurement of the result in REW without moving the microphone at all, you will find what you measure will be pretty close to the predicted curve in DL. Nobody's two ears are in exactly the same position though, plus people move their heads too...

On the timing front I'll admit I've never tried exactly what you did, in that when I've disabled the adjustment it's because I've wanted there to be no time delay applied. I'm a little surprised it behaves as you describe but thankfully it does still sound like it gives you a way to achieve what you want.

Unfortunately it is not really giving me what I want. I was hoping that a pre-set time delay on the right channel by 0.44ms to correct issues in the SHD's output settings would become a baseline and dirac doing it's thing on top of it without "correcting" my settings by disabling the delay/gain corrections by dirac. I tested with only 1 measurement to win time and indeed after exporting the filter to SHD, I get an almost perfect target curve measurement in REW (red curve in the attached image). However I can hear that the center has been shifted again to the right even though I didn't change anything on the time delay part, SHD software shows still my pre-set setting of 0.44ms and nothing is shown on the input/dirac side regarding the left channel to counter-act my settings. It seems that in some way the time delay was set back to 0 though. I altered the settings to correct by setting it again as 0.44ms, so putting the same value for the time delay which corrects the center. A new measurement (green curve) in REW however results in a much worse frequency response curve compared to the first measurement.

What I want is an almost perfect frequency response curve (red curve), with a correct centered soundstage. Now I have only a correct centered soundstage or a perfect frequency response curve

I think the dormer in my room cause a certain (perceived) delay by around 15 cm (0.44 ms) but speaker placement and speaker output are almost symmetrical.

 

@hannesb,
Just a thought but rather than mess up the time delays as you say the speakers are the same distance, let dirac do its thing and allow delay and gain changes then afterwards change the gains on the outputs tab a bit like what a balance control would do to centre your image. 

Yes I did in the past, before I found out I was able to set a time delay. However it was a difference around 3-4 db between the 2 channels and dirac corrected this too by lowering the gain on the other channel. I can try again now I know how to disable the gain/time delay in dirac. However it was audible that the right channel had lower gain settings which is not the case when changing the time delay. 

I think how the speakers are placed in a symmetrical way compared to the listening position is interpreted by dirac  as correct. Every attempt to correct in the output settings of the SHD beforehand, gets "corrected" afterwards. But the room itself does tricky things I think, with a result in an off-centered soundstage which is difficult to resolve. 

Yes but what I'm saying is set your gain and delay settings to 0.
Then run Dirac and allow it to correct timings (and gains) as it sees fit.
Then play a track with a strong centre image and gradually up the gain on one channel say by +0.5dB and reduce it on the other by -0.5dB and repeat until the centre image moves to where you want it. 

hannesb wrote:

What I want is an almost perfect frequency response curve (red curve), 

No, you really don't. That was the point I was trying to make above. Stop looking at the frequency response at a single point (where it is impossible to have both ears) and judge based on the sound. This is a totally separate issue to your delay/centering one, which is the one to focus on not the frequency response one. Also note that generally it's best to consider the frequency response of each speaker playing individually, rather than both together which I suspect you may have done above.

Just to rule one thing out, if you listen on headphones do you get a central 'image', or is if off-set to one side? Just making sure this is definitely a room rather than a hearing issue. I'm asking as I'd expect Dirac Live to help to some degree with the former but obviously wouldn't with the latter. 

 

@ asx77 I tried, but the issue is that there is a difference of over 4 db, which results in the right channel output being too low and a skewed soundstage focusing on the left and middle...

Ultrasonic wrote:

hannesb wrote:

What I want is an almost perfect frequency response curve (red curve), 

No, you really don't. That was the point I was trying to make above. Stop looking at the frequency response at a single point (where it is impossible to have both ears) and judge based on the sound. This is a totally separate issue to your delay/centering one, which is the one to focus on not the frequency response one. Also note that generally it's best to consider the frequency response of each speaker playing individually, rather than both together which I suspect you may have done above.

Just to rule one thing out, if you listen on headphones do you get a central 'image', or is if off-set to one side? Just making sure this is definitely a room rather than a hearing issue. I'm asking as I'd expect Dirac Live to help to some degree with the former but obviously wouldn't with the latter. 


 

It is definitely a room issue It is not related to my hearing, receiver (I tested several ones) or streamer (before the SHD I had a Bluesound). In every case I get the same issue  

hannesb wrote:

It is definitely a room issue It is not related to my hearing, receiver (I tested several ones) or streamer (before the SHD I had a Bluesound). In every case I get the same issue  

OK, so if you do a full Dirac measurement set, save filters, and then manually add your apparently preferred 0.44 ms delay to the required output channel, how does it sound?

Edit: just to add that hardware issues would be fixed by Dirac Live just fine (that's actually part of it's purpose), it's only hearing related issues that wouldn't be.

Ultrasonic wrote:

hannesb wrote:

It is definitely a room issue It is not related to my hearing, receiver (I tested several ones) or streamer (before the SHD I had a Bluesound). In every case I get the same issue  

OK, so if you do a full Dirac measurement set, save filters, and then manually add your apparently preferred 0.44 ms delay to the required output channel, how does it sound?

Edit: just to add that hardware issues would be fixed by Dirac Live just fine (that's actually part of it's purpose), it's only hearing related issues that wouldn't be.

A correct centered soundstage always sounds better, more natural What I want to prevent is that over the frequency of 20 - 20000 I'm having too much differences in sound levels and hence lose information in the music. This is the best system I ever had, so I don't really have a reference of how it "should" sound except for those measurement to give me some insights   

hannesb wrote:

A correct centered soundstage always sounds better, more natural  

Errr, I never suggested having a correctly centered soundstage wasn't preferable. What was the answer to my last question? I'm trying to understand what may be going on... 

For information, it's possible to use REW to measure arrival time differences between the two speakers by making use of a consistent acoustic timing reference. So if you wanted to see what Dirac or anything else is doing in this regard you can. It might actually be informative to make measurements in REW like this without DL on but with your 0.44 ms delay applied to give your central image. Make measurements for each speaker separately, and if you share the mdat file here I'll have a look. I'm thinking seeing measurements of what sounds central to you may be informative... @asx77 has made a good point re. level balance rather than time delay being important for centering by the way.