After years of frustrating failed attempts to get Dirac working properly with first 1, then 2 and now 3 subwoofers on the NAD M33, I finally achieved a balanced, natural result. I thought it was worth sharing what worked for me.
NAD M33 + Three Subwoofers + Dirac Live full
System mainly stereo listening: NAD M33 | BluOS 4.12.x | Dirac Live 3.14.1 | Cabasse Murano Alto + 3x Cabasse Santorin 30-500 2 front connected in parallell to sub out 1, next to each speakers + one in the back connected to sub ou 2, 70Hz crossover | REW | UMIK-1 | MacBook Air M3 | Official Apple USB-C to USB-A adapter (this matters, more below)
The room is the hometheater, with a large preference for stereo listening over surround, already sounding decent with some basic acoustic treatment nothing invasive. Hours of serious listening: Jazz, Classical, and whatever else takes my mood.
Quick summary (dedicated listening room with one MLP)
- Align subs with REW first, timing, phase, level and note the volume knob positions. I find this did give a better foundation for dirac to work with.
- In Dirac's volume calibration screen, reduce sub volume controls until subs roughly match mains, use only volume on subs, not dirac individual volume control
- In the same screen, reduce the louder main speaker by the measured difference to help centre the image
- Wait for a quiet moment Dirac needs a low noise floor to measure the reduced-level subs, this can be quite difficult in a busy street like where I live, master gain at -40db usually works for me.
- Official Apple USB-C adapter with UMIK-1, on Mac I did find out third-party adapters corrupt HF measurements in my case.
- 0-degree mic orientation facing speakers
- Run Dirac calibration: Single measurement at MLP for focused listening, then maybe add more if not satisfying with only one.
- After filter export, return sub volume controls to the REW reference positions
The first calibration using all of these together produced the best result in five years. Correct bass integration, immediately recognisable balanced and natural sound something I had never achieved with Dirac before. On large-scale classical music the stereo image is pinpoint precise, all instruments in tonal balance, and the coherence is simply better than Dirac off which was already quite decent. Don’t touch it anymore ;-)
Dirac off:
View attachment 521155
Dirac on with automatic filter:
View attachment 521153
The frustrations
The image kept drifting left. Even adding 1–2dB balance correction after calibration never fully fixed it.
With Dirac on, it sounds unengaging. Without Dirac it sounds lively and engaging but the bass is messy. Using Dirac only for bass correction shows a massive gap in the spectrogram between mains and subs where did the drum slam go? Where did the bass go?
The M33's subwoofer level handling in Dirac is just broken in my opinion. Whatever I tried adjusting the filter, disabling delay/gain, tweaking the master output, Dirac consistently crushed the subs relative to the mains. I gave up more than once.
Subs crushed whatever I tried:
View attachment 521157
Disabling delay / gain just makes it worse:
View attachment 521158
What finally worked was not one tip gathered on ASR, it was applying all of them at the same time.
What worked
Starting with a decent sounding room and good speaker/sub placement as much as possible.
Aligning subs with REW before touching Dirac
Use REW to align the subwoofers to the mains as much as possible: timing, phase, and level. Note the exact position of every volume knob on every sub. This is your reference level for real listening. Do not skip this. I’ve used a wireless sub link for my rear subwoofer, the delay added compared to wired link just made things worse and more difficult to align timing of all subs combined. Back to wired.
Reducing sub volume for Dirac calibration
This is the key workaround for the Dirac / M33's broken sub level handling. Whatever I tried dirac killed the bass, when level matching subs with REW, subs show as 10 db louder in dirac volume matching step in recent versions, very annoying. Bass are killed also when matching sub level in dirac, reducing volume of subs in dirac by 10 db brings even worse results.
Physically turn down the sub volume controls until the subs roughly match the mains level during Dirac's volume calibration screen.
The annoying catch: the room needs to be quiet enough for Dirac to measure the subs at this reduced level. With recent Dirac versions the noise floor requirement with reduce subwoofer volume is frustratingly strict. I had to wait for a day when ambient noise was naturally low in my case
Fixing the image shift
In Dirac's level calibration screen, reduce the level of whichever main speaker measures higher to match the other. This is what finally gave me a stable, centered image possibly combined with a very similar mains speaker response after calibration. It is clearly described in the manual. I wish I had tried it years earlier, I thought adjusting balance would have the same effect.
Step 4 Run Dirac calibration
Nothing special here. Just run it normally, for the first time I did take a REW measurement to see what happened after the first measurement and I was shocked by the result.
Step 5 Return subs to REW reference level
After the Dirac filter is exported and loaded onto the M33, return all sub volume controls to the original REW-aligned positions. The subs are now at the correct listening level with Dirac's correction applied.
Other things that seem to make a real difference
The USB adapter seem to matter
I use a UMIK-1 on Mac. Switching from various third-party USB-C adapters to an official Apple USB-C to USB-A adapter immediately produced consistent, believable high-frequency measurements. Previous sessions with other adapters gave erratic HF results. If your high-frequency REW measurements look inconsistent between sessions and nothing acoustic has changed, check your adapter.
Old measurement with typical high frequency weirdness, sometimes a complete drop from 10k or a raise like below:
View attachment 521159
Microphone orientation: 0 degrees facing the speakers
The recommended 90-degree ceiling position produces slightly lifeless result for me. Switching to 0 degrees facing the speakers gives back some of the lost life in the music. Worth trying both it is easy to switch and re-measure.
Highly controversial: single measurement at MLP
One measurement at the exact listening position gave me the best result I have ever achieved, significantly better than stadard 9+ multi-point calibration attempt over the years. Floor bounce and dips above 100Hz were corrected precisely for the first time.
Specific to Murano alto: Speaker ports open
Running the Muranos with ports open gives a tiny bit more dynamic/ punchy result. More alive. Subs take over the frequency of the port tuning anyway with a 70 HZ crossover.
During my experiments with accoustic pannels or bass traps behind the muranos I did find that it does suck the life away from them.
www.superbestaudiofriends.org
