Dirac Research Questionnaire

[hearhere]

I received by email today a questionnaire from Dirac Research - I expect many here have also received this. It has given me an opportunity to explain exactly why I believe (and confirmed by simply listening carefully) that Dirac and its competitors, when built into full-range amps, are bad for ultimate sound quality.

I've expressed my views here on this and other forums and explained why this spoiling of ultimate sound quality may be. No one has challenged my theory, but many have said that their own systems have benefited from RC DSP.

I beg anyone who is using Dirac user (or RoomPerfect or MARS or any other RC DSP built into full-range amps and who has first class speakers) to do this simple test. Listen to exciting music with not much bass (whatever should be giving you goose-bumps) with the filter engaged and then with the filter not engaged and back again. Try to ignore the bass (the area where RC adjusts for poor room acoustics) and concentrate solely on the top end - where the goose-bump factor is generated - and see what you think.

I've tried Room Perfect, Dirac and MARS and all have damaged the important top end of the frequency range just enough to reduce or to lose this goose-bump factor, even though the processor doesn't ADJUST the signal above the bass frequencies - 500 Hz in my version of Dirac Live.

I don't disagree that the job of adjusting for poor room acoustics can be simplified by RC DSP and that those with multi-speaker systems (2 channel plus subs, or AV systems) may find conventional tuning very difficult and RC DSP makes the job simple with an apparent immediate "improvement" in sound, but critical listening will (in my experience as well as others) reveal that the top end is damaged by this extra (and totally unnecessary for high frequencies) stage of signal processing, in a similar way that tone controls and graphic equalisers of the past used to. Any extra signal processing is fundamentally a bad thing!

If (and only if) the frequency range is spilt (by active crossover) first and the bass only is sent to the RC DSP and then onto the bass amp and driver, while the higher frequencies continue unmolested by DSP to their own amp and drivers can this problem be avoided. This is probably why those few speakers on the market that have amps and DSP inside their cabinets sound so good. They perform room correction the “proper” way by processing only the bass. Speakers that do this include Dutch & Dutch, Kii and Avantgarde XD range. More will follow. Perhaps the KEF LS60 (the subject of another current thread) also does – I’ve not investigated.

I’d invite anyone with first-class speakers capably of generating goose-bumps for itheir listener to do the test I’ve described and respond with your unbiased and subjective findings. You may well think on balance that RC DSP improves the overall sound, but does it spoil the top end in the process? Thanks. Peter

 

[tuga]

I received by email today a questionnaire from Dirac Research - I expect many here have also received this. It has given me an opportunity to explain exactly why I believe (and confirmed by simply listening carefully) that Dirac and its competitors, when built into full-range amps, are bad for ultimate sound quality.

What do you mean by "built into full-range amps"?

Are you referring to applying DRC/EQ to the whole signal or processing the whole signal?

 

[tuga]

With any OS (oversampling/upsampling) DAC the whole signal is processed before conversion, that is why I think that your post needs clarification.
The same is true if one uses digital volume control.

 

[tuga]

As a side note, the problem I have with auto correction devices is that even though they require very little learning effort and interaction from the user we are left at the mercy of whatever the automation has been programmed to produce.
We trade comfort and ease of use for effectiveness.

 

[hearhere]

What do you mean by "built into full-range amps"?

Are you referring to applying DRC/EQ to the whole signal or processing the whole signal?

Any full-range amp (such as NAD, Lyngdorf, Micromega) that has built-in RC requires the entire frequency range to pass through the filter if it's engaged. So my NAD M33 has Dirac built in. There is no way the higher frequencies can not pass through the processor if the filter is engaged. Only if the signal is first split into bass and non-bass and the bass alone goes through DSP can the non-bass bypass the filter. I hope that makes sense and is clear. Peter

 

[hearhere]

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

 

[tuga]

Any full-range amp (such as NAD, Lyngdorf, Micromega) that has built-in RC requires the entire frequency range to pass through the filter if it's engaged. So my NAD M33 has Dirac built in. There is no way the higher frequencies can not pass through the processor if the filter is engaged. Only if the signal is first split into bass and non-bass and the bass alone goes through DSP can the non-bass bypass the filter. I hope that makes sense and is clear. Peter

Apparently you have to pay for full-range Dirac with the M33:

Your NAD M33 model comes with Dirac Room Correction Limited Bandwidth installed, giving you the ability to address the most common acoustic problems your listening room might have, such as muddy or booming bass. However, to truly enjoy the full quality of your new receiver, we strongly recommend that you invest in the Dirac Live Room Correction Full Bandwidth.

Dirac Live Room Correction tiers ​

Dirac Live adds value to any sound system, be it an entry-level soundbar, a high-end home theater system, or a professional recording studio. To offer our business partners flexibility for licensing Dirac Live, we’ve developed three tiers.

— Ready: The system is prepared for Dirac Live Room Correction. Just purchase an end-user license to activate it.
— Limited bandwidth: The system can perform room correction up to 500 Hz, and the target curve can be tailored within this range. This correction is sufficient to deal with the bass and voice areas but not the entire audible spectrum. Upgrade Limited bandwidth to full bandwidth.
— Full bandwidth: This is the top tier of the room correction feature. It allows the system to be corrected across the entire frequency range while giving you full control of the target curve.

 

[tuga]

Do the higher frequencies still sound bad when you disable Dirac on your M33?

 

[tuga]

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

I am not against your idea in principle but my experience is that when DRC/EQ is applied in the upsampling step with high bit-depth processing there are no artefacts added to the sound (as it happens with HQPlayer).

One of the effects of room correction (digital or analogue) is a perceived drop in bass level ("less", "cleaner", "dryer" bass) and that in turn results in a perceived increase in the upper-mids and treble.
Since Dirac allows for custom target curves, maybe you could try a shelving filter of some 3dB in the bass.

 

[tuga]

Having a processor inside the DAC/amp case is probably not the best idea since when doing heavy computational weight-lifting (such as DRC) it will generate noise which could affect the D/A conversion and have audible effects in the treble.
And Stereophile's M33 measurements don't show a particularly clean or low noise-floor to start with.
All this could be exaggerated by the high sensitivity of your horns.

In other words, the problem might not be Dirac per se but the way it's been implemented in the NAD.

 

[tuga]

If (and only if) the frequency range is spilt (by active crossover) first and the bass only is sent to the RC DSP and then onto the bass amp and driver, while the higher frequencies continue unmolested by DSP to their own amp and drivers can this problem be avoided. This is probably why those few speakers on the market that have amps and DSP inside their cabinets sound so good. They perform room correction the “proper” way by processing only the bass. Speakers that do this include Dutch & Dutch, Kii and Avantgarde XD range. More will follow. Perhaps the KEF LS60 (the subject of another current thread) also does – I’ve not investigated.

I am sure that both the Dutch & Dutch and the Kii process the whole signal because they are using digital crossovers, and performing (driver) EQ and impulse response correction.

 

[hearhere]

Do the higher frequencies still sound bad when you disable Dirac on your M33?

No - As I said, it's when the filter is engaged that the ultimate sound quality is reduced. With No Filter the top end sparkles more and this contributes to what I describe as the goose-bump or tingle factor that one experiences at live performances - if they are good performances!

In other words, the problem might not be Dirac per se but the way it's been implemented in the NAD.

No, again as I mentioned I've experienced the same when I had a Lyngdorf TDAI-3400 amp (with RoomPerfect) on loan and with my Micromega M100 (with MARS). Listening carefully to all these syatems, the top end sparkle was reduced with the filter engaged.

Apparently you have to pay for full-range Dirac with the M33:

Yes but that doesn't change the argument, The fact is, whichever version of Dirac one has, if it's built into a full-range amp, the entire frequency range has to pass through the processor, despite the levels of higher frequencies not being adjuted in the version in my NAD.

I am sure that both the Dutch & Dutch and the Kii process the whole signal because they are using digital crossovers,

I don't see why that should be. It's because D&D use digital crossovers that they can bypass the top end around the room correction processor. Why shouldn't they do that, as the room acoustics don't (to any significant degree) upset the top end - it's the bass that needs to be corrected. But as I've claimed elsewhere, even this processing should be the very last resort if all other set up procedures fail.

 

[tuga]

No, again as I mentioned I've experienced the same when I had a Lyngdorf TDAI-3400 amp (with RoomPerfect) on loan and with my Micromega M100 (with MARS). Listening carefully to all these syatems, the top end sparkle was reduced with the filter engaged.

What makes you believe that DRC is not poorly implemented in those other amplifiers?

 

[tuga]

I don't see why that should be. It's because D&D use digital crossovers that they can bypass the top end around the room correction processor. Why shouldn't they do that, as the room acoustics don't (to any significant degree) upset the top end - it's the bass that needs to be corrected. But as I've claimed elsewhere, even this processing should be the very last resort if all other set up procedures fail.

In a digital crossover the whole signal is processed. Crossovers don't just high-, low- or band-pass, they also EQ drivers (to flatten response and or the impedance).

Filters, EQ, oversampling are just math. That math may be lossless or lossy depending on the bit-depth and sample-rate of the processing. And it may generate noise which can affect the D/A conversion and impact the treble.
It is why some people go to great lengths to isolate network streaming and DSP processing from the DACs.
All-in-ones in my view are not ideal. Convenient, yes, but crippled by default.

I think that the best way to confirm or disprove your theory would be to use off-board Dirac (in a computer).

 

[hearhere]

What makes you believe that DRC is not poorly implemented in those other amplifiers?

I'm absolutely not saying that. Any signal processor is basically bad new for the signal however well it may be implemented. It's not poor implementation, it's just an unnecessary extra processing stage of the top-end signal that doesn't need to be processed.

 

[tuga]

I'm absolutely not saying that. Any signal processor is basically bad new for the signal however well it may be implemented. It's not poor implementation, it's just an unnecessary extra processing stage of the top-end signal that doesn't need to be processed.

I disagree.
Unless you use a NOS DAC the signal will go through a processor, and the reason why over- or up-sampling and filtering is used is to improve the reconstruction of the top-end signal.
If it sounds bad, it is very likely down to implementation.

There's been a recent discussion at the REG forum about the Lyngdorf sounding bad, this from a guy to which 00s and 11s all sound the same. And it's not just because the Lyngdorf applies EQ across all the audio range but due to audible artifacts.

I am using off-boad processing (upconversion of Redbook PCM to DSD, noise-shaping, filtering and EQ'ing) and the sound is audibly "smooth" or "effortless".

 

[Blzebub]

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.

 

[hearhere]

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.

Yes, I agree and there are plenty of other ways to mitigate these room acoustic problems that are better than chucking RC DSP at them. Precise speaker placement, toe-in and maybe tilt, plus careful consideration of carpeting, curtains, furnishings will generally achieve the object. The less signal processing, the better for the signal and hence the sound we hear from our speakers.!

 

[hearhere]

I hope someone will sometime do the simple test I suggested in my opening post, rather than bring red herrings into the discussion! The points raised so far, with rspect, are covered in my top post and the object was to invite members to critically listen to exctiting, bass-shy music through their own high quality speakers with and without their DSP filter. In my own case this is dead easy - there's a Setting in the BulOS control app that requires a single click to move from Dirac Filter to No Filter. Thanks. Peter

 

[chris]

I use Dirac and have setup filters across the full frequency spectrum as well as only up to 200Hz. For well recorded tracks without too much low bass I agree with you that it does sound better with Dirac off. This is more noticeable with full range filters. Vocals in particular seem to be a little more forward and there is more zest to the music.

Since most of us have less than perfect listening rooms and can’t fill them bass traps, there is usually a problem with room modes. I have spent a long time trying different speakers in different positions with different listening positions and although that can make a big difference it can only take you so far. In my room, with my system and my ears a lot of music sounds better with Dirac on than off. I have filters which act across the full frequency range so that i can tame overly bright recordings as well as low frequency only filters to reduce the room modes and lift the bass a little. It is easy to switch between the filters and turn it off entirely,

So I did complete the questionnaire and said that I found the software easy to use, it improved a lot of tracks, but wished I did not need it.