Dirac Research Questionnaire

[tuga]

Any form of DSP/EQ mangles the original signal. Room modes don't go away because you have EQ. All that happens is that the room modes are excited to a lesser extent, but the temporal overhang of the modes persists. It's a terrible idea.

Have you seen what happens to the temporal overhang when one brings down the level of room-generated peaks?
DRC doesn't solve room issues but it does improve the frequency response in the bass and sub-bass at the listening spot.

Any OverSampling CD player or DAC will run the signal through DSP. It mangles the sound and what comes out is better or more accurate than what goes in.

Why is up- or over-sampling important?

• To improve the effectiveness of the brick-wall filter, indispensable to correctly reconstruct the original signal and avoid aliasing.
• To reduce jitter (oversampling reduces jitter noise by 3 dB for every doubling of the sampling rate).
• To improve S/N ratio.
• To feed the D/A chip with the highest admissible sample rate and/or the optimal rate from a design perspective (most current D/A chips will take at least 24-bit/384kHz). Benchmark currently up- or down-samples all its inputs to 211kHz for the ESS chip (reasoning explained here).

Archimago objectively demonstrates the effectiveness of Upsampling and Upconverting through the use of HQPlayer here.

 

[Blzebub]

Have you seen what happens to the temporal overhang when one brings down the level of room-generated peaks?
DRC doesn't solve room issues but it does improve the frequency response in the bass and sub-bass at the listening spot.

What happens to the peak transient level? There is no such thing as a free lunch.

 

[SteveS1]

I've used the stand alone Dirac for a long time. Initial results were overall a bit underwhelming but some aspects were definitely helpful and encouraged me to persist, so I carried on fine tuning and experimenting with measuring and filter creation. Given the amount of faffing around with kit and speaker/listening positioning I had done to minimise room issues in the past, it really was no big deal to apply a bit of time and effort.

With the latest Dirac Live Processor and DiracLive3 App there is an excellent optimised filter creation process using your room measurements and it is a much better starting option than the old default target curve and requires mimimal alteration if any (which is still perfectly possible should you wish). I wouldn't be without it, it's a significantly more enjoyable sound - switching it out is sobering and the physical measures to get anything like as good a result would be a non-runner in my room.

 

[bencat]

I have only used Dirac as a stand alone DSP in both active and passive systems . This method has complete control of exactly were you add ot do not add any filter and you can set what it changes and what it does not . In my main music system which is passive the difference is very much an improvement . While it is no as easy I did try the test for Dirac on and off and preference was most certainly for on and in my case full spectrum . As @SteveS1 says the latest filter and settings are much improved over Dirac 1.0 and as with all Dirac you have complete control over any filter which you can change on the fly while listening to music. I make no comment on other DSP software as I have never used them . For me DIRAC Live is helping me to get the best music I can in my room and I would not be without it . Others with their music and their ears may well feel different .

 

[karlsushi]

I have posted my experiences on the Wam elsewhere, but in summary, after initially being wowed by what Dirac could do with bass issues that plague my living room system, I too have experienced an added digital glare to the mid and especially treble, with Dirac turned on. This is even when only adding a filter (curtained) for frequencies below 180Hz (where my issues are).

I use the PC version direct from a Windows Laptop via J:River.

Out of interest, I emailed Terry at Pursuit Perfect System the other day about it. He is a professional Dirac installer and uses it as part of all of his reviews, including many very high end units and speakers.

Whilst he obviously couldn't provide a free consultation service, he was very helpful. He suggested that the digital glare I experience shouldn't be happening and felt that this was perhaps an issue somewhere else within my system, with his main finger pointed at my choice of a laptop as a source.

What is interesting is that the digital glare is completely removed when turning off Dirac, so either Dirac is the issue, or it is something else going on with my laptop's ability to process Dirac.

Whilst I haven't given up on it completely, I am tending towards the idea of moving to a system with stand-mount speakers, with separate subs with internal DSP e.g. Martin Logan subs or similar.

It is a shame because music with floor-standers in that room is now unlistenable with and without Dirac. The former because of the glare and the latter because of the boomy bass.

 

[Tony_J]

I have only used Dirac as a stand alone DSP in both active and passive systems . This method has complete control of exactly were you add ot do not add any filter and you can set what it changes and what it does not . In my main music system which is passive the difference is very much an improvement . While it is no as easy I did try the test for Dirac on and off and preference was most certainly for on and in my case full spectrum . As @SteveS1 says the latest filter and settings are much improved over Dirac 1.0 and as with all Dirac you have complete control over any filter which you can change on the fly while listening to music. I make no comment on other DSP software as I have never used them . For me DIRAC Live is helping me to get the best music I can in my room and I would not be without it . Others with their music and their ears may well feel different .

I have had similar experiences with my various DIY active systems. I use Dirac full range and for me, the performance with Dirac enabled is a significant improvement on the performance without. That difference has become smaller as my ability to match the levels of the drivers and adjust crossover points, etc. have improved, but the application of Dirac still significantly improves my listening experience.

I've yet to experience/recognize this "digital glare" of which others have spoken - not at all clear to me either what it is or whether it is the DSP/digital signal's fault.

 

[Tony_J]

Let me try to illustrate with a diagram to show what I mean by allowing the Non-Bass to bypass the RC DSP.

The box in your diagram labelled Active Crossover is of course a DSP-based crossover, so in both of your options, the entire frequency range is subject to DSP processing. There's nothing magical about RC vs. DSP in general - at the end of the day, all that is going on is the application of digital filters. The only "magic" is in how you arrive at the specification of those filters. In the case of Dirac, it does so by taking various measurements and then deciding what to specify; in the case of a (simple) digital crossover where all you are doing is splitting the digital signal, the user defines a crossover frequency and crossover curve. The end result is the same in the sense that the digital signal, in its entireity, is processed according to those filters to produce the end result. I don't doubt that you are hearing a difference with Dirac enabled vs disabled, and that you prefer it to be disabled, but don't imagine that the difference is because only part of the signal is being DSP processed - that just isn't the case in the two alternatives that you have illustrated.

If you want to create a setup where the non-bass isn't DSP processed (beyond any DSP that occurs in the DACs), then the lower half of the diagram would look a little different. You would need to disconnect the "DAC and non-Bass Amp" box from the (digital) Active Crossover, connect the Non-bass DAC direct to the Digital Source (same digital signal applied to the Active Crossover and the non-bass DAC), and insert another (analogue) crossover between the non-bass DAC and the non-bass amp. That way, the non-bass is split off from the bass without any additional DSP being applied. However, going that route would have its own issues, because there will be a measurable delay introduced in the bass (DSP) path relative to the non-bass (analogue) path due to the insertion of the DSP in the path, so it would be necessary to insert a similar delay in the analogue path to compensate (or adjust the positioning of the drivers to delay the non-bass relative to the bass), if you were wanting to truly compare the two approaches.

 

[hearhere]

Have you seen what happens to the temporal overhang when one brings down the level of room-generated peaks?
DRC doesn't solve room issues but it does improve the frequency response in the bass and sub-bass at the listening spot.

Any OverSampling CD player or DAC will run the signal through DSP. It mangles the sound and what comes out is better or more accurate than what goes in.

Why is up- or over-sampling important?

• To improve the effectiveness of the brick-wall filter, indispensable to correctly reconstruct the original signal and avoid aliasing.
• To reduce jitter (oversampling reduces jitter noise by 3 dB for every doubling of the sampling rate).
• To improve S/N ratio.
• To feed the D/A chip with the highest admissible sample rate and/or the optimal rate from a design perspective (most current D/A chips will take at least 24-bit/384kHz). Benchmark currently up- or down-samples all its inputs to 211kHz for the ESS chip (reasoning explained here).

Archimago objectively demonstrates the effectiveness of Upsampling and Upconverting through the use of HQPlayer here.

Tuga - Thanks but I can't help feeling you are more concerned with theory and measurements than how the music actually sounds at one's listening chair. I admire your ability to take measurements and create graphs because I simply follow the Dirac or RP instructions and have no way of checking the results apart from sitting down and switching between filter and no filter.

Having said that when I moved to this new-built flat, the acoustics were dire with reverberation being an obvious problem. Over time, I added carpeting to 30% of the floor area, hung curtains (though never closed, they have some beneficial effect), added more soft furnishings and continually adjust speaker positions, toe-in, etc. I think I've achieved pretty good sound to the extent I don't believe I need room treatment panels (I wouldn't be prepared to spoil my room with them anyway) and the RC DSP filter is not engaged for best (highest unmeasurable goosebump factor) sound. The intention of my original post here was to ask those with 2 channel systems (preferably no subs) to do the simple test I suggested with suitable music and to report objectively their findings. A simple non-technical listening test that doesn’t require theory or measurements – only ears needed. Peter

 

[hearhere]

in the case of a (simple) digital crossover where all you are doing is splitting the digital signal, the user defines a crossover frequency and crossover curve.

That's precisely what I intended to suggest in my box marked Active Crossover. It doesn't include the room correction DSP - just splits the signal into bass and non-bass.

Let me mull over your last paragraph later. I thought my diagram was pretty clear on how the various components should work, although some would be within a shared enclosure. Peter

 

[Tony_J]

That's precisely what I intended to suggest in my box marked Active Crossover. It doesn't include the room correction DSP - just splits the signal into bass and non-bass.

My point is that your throwaway phrase "...just splits the signal into bass and non-bass" obscures the fact that in order to achieve that split, the DSP is actually processing the entirety of the digital signal, not just those frequencies close to the crossover point.

 

[SteveS1]

I've yet to experience/recognize this "digital glare" of which others have spoken - not at all clear to me either what it is or whether it is the DSP/digital signal's fault.

Early on my concern was the opposite, in that Dirac was dealing with the low end very well but taking too much of the 'life' out of the mids and 'over restraining' things - for want of a better description. As I experimented, I found I was able to retain the improvements without compromise, mainly by making small adjustments to the target curve and was very happy with it. The newer version definitely helps by producing an optimum filter that I find difficuIt to fault, so I've not been tempted to fiddle.

I've not experienced 'glare' or over brightness with it at any stage.

 

[tuga]

What happens to the peak transient level? There is no such thing as a free lunch.

The room increases the amplitude or level, EQ decreases that level. EQ doesn't affect the transient (it only acts on the level which has been exaggerated by the room), the room does.

 

[tuga]

Whilst he obviously couldn't provide a free consultation service, he was very helpful. He suggested that the digital glare I experience shouldn't be happening and felt that this was perhaps an issue somewhere else within my system, with his main finger pointed at my choice of a laptop as a source.

That is my view also, as I wrote earlier in the thread.

 

[tuga]

Tuga - Thanks but I can't help feeling you are more concerned with theory and measurements than how the music actually sounds at one's listening chair. I admire your ability to take measurements and create graphs because I simply follow the Dirac or RP instructions and have no way of checking the results apart from sitting down and switching between filter and no filter.

Having said that when I moved to this new-built flat, the acoustics were dire with reverberation being an obvious problem. Over time, I added carpeting to 30% of the floor area, hung curtains (though never closed, they have some beneficial effect), added more soft furnishings and continually adjust speaker positions, toe-in, etc. I think I've achieved pretty good sound to the extent I don't believe I need room treatment panels (I wouldn't be prepared to spoil my room with them anyway) and the RC DSP filter is not engaged for best (highest unmeasurable goosebump factor) sound. The intention of my original post here was to ask those with 2 channel systems (preferably no subs) to do the simple test I suggested with suitable music and to report objectively their findings. A simple non-technical listening test that doesn’t require theory or measurements – only ears needed. Peter

You've caught me red-handed. I am only concerned about the theory and measurements... I don't even listen to music.

I am not doubting that Dirac could be having an audible negative impact in your system, I only doubt your theories about the potential causes. Maybe to your limited understanding of the theory (not that mine is that much greater to be fair) is leading you in the wrong direction.

Unlike @Tony_J I have experienced "glare", "hardness" and "grain" with some digital sources; the causes are varied and not easy to track down. I have listed a few earlier in the thread.

 

[Blzebub]

The room increases the amplitude or level, EQ decreases that level. EQ doesn't affect the transient (it only acts on the level which has been exaggerated by the room), the room does.

EQ decreases the level, correct. EQ can not discriminate transients, it just cuts the frequencies you ask it to.

So, what effect does that have on the direct sound (as opposed to the reverberent sound), and which is more important to the listener, direct or reverberent?

 

[Blzebub]

I don't know what digital glare is either, but I suspect it's imaginary.

 

[hearhere]

You've caught me red-handed. I am only concerned about the theory and measurements... I don't even listen to music.

I am not doubting that Dirac could be having an audible negative impact in your system, I only doubt your theories about the potential causes. Maybe to your limited understanding of the theory (not that mine is that much greater to be fair) is leading you in the wrong direction.

Unlike @Tony_J I have experienced "glare", "hardness" and "grain" with some digital sources; the causes are varied and not easy to track down. I have listed a few earlier in the thread.

am not doubting that Dirac could be having an audible negative impact in your system, I only doubt your theories about the potential causes. Maybe to your limited understanding of the theory (not that mine is that much greater to be fair) is leading you in the wrong direction.

I'm wondering if you have a room correction DSP in your own system. I'm not familiar with much of your equipment per your Profile, apart from the LS3/5A speakers, but is there an RC DSP there? If not, then you can't do the simple "with or without DSP filter" test I'm asking for.

The simple fact is that I have found that RC DSPs take a little of the "life" (as another contribulor here describes it) out of the music and I have concluded that it's because the entire frequency signal has to be subjected to this complex DSP. The top end (ie non-bass) doesn't need to be adjusted, yet it can't help having to go through the DSP unless done by the way shown in my sketch.

If you can do the test I suggest with the type of music I describe, I'd be very happy to hear your subjecteive description of the With and Without filter sound listening only to the top end where the "life" of the music is mainly delivered. Thanks. Peter

 

[tuga]

I'm wondering if you have a room correction DSP in your own system. I'm not familiar with much of your equipment per your Profile, apart from the LS3/5A speakers, but is there an RC DSP there? If not, then you can't do the simple "with or without DSP filter" test I'm asking for.

The simple fact is that I have found that RC DSPs take a little of the "life" (as another contribulor here describes it) out of the music and I have concluded that it's because the entire frequency signal has to be subjected to this complex DSP. The top end (ie non-bass) doesn't need to be adjusted, yet it can't help having to go through the DSP unless done by the way shown in my sketch.

If you can do the test I suggest with the type of music I describe, I'd be very happy to hear your subjecteive description of the With and Without filter sound listening only to the top end where the "life" of the music is mainly delivered. Thanks. Peter

My speakers are not the LS3/5As (left stand) but the slightly larger LS3/6s (right stand):

As mentioned in a previous message, my playback system uses DSP (HQPlayer) to upconvert Redbook PCM to DSD256 and DRC (custom set P-EQ filters to address a couple of room-generated bass peaks).
Turning RC EQ filters on or off only has no audible impact on sound except for the frequencies affected by the filters, and it's a good impact.

Left speaker
Filter 1: ON PK Fc 49.55 Hz Gain -9.90 dB Q 7.500

Right speaker
Filter 1: ON PK Fc 48.10 Hz Gain -9.00 dB Q 6.750
Filter 2: ON PK Fc 136.4 Hz Gain -4.00 dB Q 7.500

 

[tuga]

I don't know what digital glare is either, but I suspect it's imaginary.

According to Stereophile's glossary "glare" refers to an unpleasant quality of hardness or brightness, due to excessive low- or mid-treble energy. Have you never experienced it with digital sources?

 

[tuga]

EQ decreases the level, correct. EQ can not discriminate transients, it just cuts the frequencies you ask it to.

So, what effect does that have on the direct sound (as opposed to the reverberent sound), and which is more important to the listener, direct or reverberent?

The room generates a peak in the frequency response, an agnostic peak which cares not about whether the sound is steady state or a transient and which takes longer to decay than the rest of the spectrum (rings or resonates).
By reducing this peak you are also reducing the time it takes for the sound at that frequency to settle.
You EQ for a precise location in the room, the place where you put the mic.

In regard to your question, loudspeakers are omnidirectional in the bass so the direct sound becomes more relevant above the Schroder frequency, which is why one should only EQ the in-room response up to ~400-500Hz and the anechoic speaker response above that: