With little understanding of how all the RoomFit settings worked, it proved to be impossible to achieve any change that I could feel comfortable with, either suspecting the result was unnatural or inaccurate.
Running RoomFit multiple times with my Android phone's microphone created quite varying results. This undermined my confidence in the tool.
Yes, the default settings are IMHO far from ideal, and this is not helped with clumsy parameter names and descriptions, and in general not very accurate documentation. There's definitely room for improvement in all these areas.
But IME another really crucial part of the puzzle needed for any room correction (including RoomFit) to work well is to first optimize speaker/sub placement and sub crossover to avoid/minimize as many nulls/dips in the response as possible.
No room correction tool can work miracles, and so it is really important to ensure as good as possible a baseline for EQ before running RoomFit.
I provide some placement advice and explanations in
this post. You can see an example of the placement optimization process in
this recent thread.
1. It's generally recommended to focus on the lower and mid frequencies where room acoustics have the most significant impact. Room modes, standing waves, and bass resonances typically occur below **500 Hz**, making this range critical for correction. Peaks and nulls in the bass region can cause muddiness, boominess, or uneven bass response.
Agreed! Note that often it is enough to limit correction even lower than that, e.g. below 300Hz or even only 200Hz.
2. Be cautious with high-frequency correction, as excessive tweaking can lead to unnatural sound.
High frequency correction (>1kHz) should in principle never be done based on steady-state in-room response measurements.
Ideally this part of the spectrum should only be "corrected" based on full anechoic loudspeaker measurements (see spinorama.org for many examples of such measurements and corresponding EQ corrections).
With the recent separation of RoomFit and EQ we're now able to load these correction profiles in EQ (as PEQ) and run them together with RoomFit! This is the ideal approach - though of course well-designed loudspeakers don't really need any response correction above the bass region, but all systems benefits from room correction in the bass!
3. Flat target curves can sound a bit dull, speaker designers themselves often aim for something more like B&K and Harmon, so try those if you want an improved listening experience.
With typical front-firing loudspeakers the "flat" target works only in extreme nearfield or in very dead rooms.
Otherwise a target with at least some downward slope is needed. The amount of sloping-down depends on specific loudspeaker directivity, room reflectivity and personal preference. I personally like to use -0,6dB/oct in my main system.
4. Many bookshelf speakers and even some floor-standing speakers and subwoofers simply cannot reproduce 20Hz in any meaningful way. They experience a sharp roll-off (a steep decline in output) starting much higher, say at 30Hz or 40Hz. Correcting rolloff will amplify noise and potentially distort following corrections. A roll-off is quite visible in the RoomFit measurements, having identified where it starts, move the bottom cutoff above this region, perhaps around 50-60Hz.
Correct!
Note that when using the new "no boost mode" this is probably not needed, as RoomFit won't try to boost the natural bass roll-off in this mode of operation.
5.Avoid Over-Correcting High Frequencies (Above 4000 Hz). High frequencies (e.g., above 4000 Hz) are less affected by room modes and more influenced by speaker design and direct sound. Over-correction here can result in a harsh or artificial sound.
Correct. Same is valid as for my comment on your point 2.
6. In Non-Boost Mode, RoomFit only reduces peaks, avoiding boosts that can strain speakers or cause distortion. This is safer and often more effective for bass correction .
This ties in to the beginning of my post. If placement and crossover is optimized there should be no need to use boost in RoomFit anyway.
In addition EQ boosts bring various potential issues - see
this post for some explanations.
7. Use a Low Q Factor. Using a Q factor like 3 or 4 ensures the filters are broad and smooth, correcting general trends rather than trying to surgically pinpoint narrow peaks and dips that may be caused by unreliable measurement data.
There's a bit more to it than this, but it is difficult to summarize all of the nuance in just a few sentences.
So I'll just say that in most cases the following is a good rule of thumb:
- High-Q (sharp) positive gain (boost) PEQ filters should indeed be avoided. But remember that we anyway want to avoid positive gain PEQ altogether when doing room correction.
- It is OK to use high-Q (sharp) negative gain PEQ filters at low frequencies (i.e. where the room dominates). This means that in RoomFit "no-boost mode" it is OK to use a high max Q value, especially if you limit the correction to low frequencies only.
Remember that room response peaks can shift under some cases, which means that the same RoomFit correction profile anyway may not work as well in all cases.
Here's a reference you might find interesting.
There's more to this story, but I hope the above will be useful as a start!
So I tried setting the cutoff range to 50Hz-500Hz, Max Q 3.5, and set Non-Boost Mode. The result was quite subtle. Some of the base seems tighter, maybe a bit less warmth, but more detail. My system still has its characteristic sound that I'm used to, which is a good thing.
Good that you found some values you're happy with! These are the values I use:
- Type: Individual Channel RoomFit
- Target: B&K
- Freq: 20Hz - 8000Hz (but in my case correction actually applies only below 500Hz, as you will see if you check this link)
- Max Gain: +6dB
- Min Gain: -12dB
- Max Q: 10
- Non-Boost Mode: enabled
- Subwoofer Calibration: enabled (Sub crossover is 80Hz, and mains are delayed by 2ms)
- Smoothing: 1/12 Octave
- Moving Mic Measurement (beta): disabled
- Multiple Measurement (beta): disabled
- External microphone: miniDSP UMIK-1 with a calibration file
Note that I've spent time optimizing the subwoofer level, crossover, mains delay and speaker/sub placement before doing room correction. The exact settings I used will not apply equally well to every system, so that should be taken into account!
In
this post you can see the response I was able to get with the above values. This gives me very smooth sounding bass, without any audible resonances/boominess or 'holes'. Of course the loudspeaker character is not lost - quite the contrary, without distracting resonances or bass suck-outs the quality of the speakers can shine through.
Hope you find some of these comments useful!
I'd be interested in others picking this appart and adding more wisdom.
Let me say at the end that IMHO you made a very good summary in this post so you obviously did a very good job researching the topic.
Thanks for taking the time to share your findings with all of us in the community!
Most of all - good luck and have fun!
.
