Fighting room resonances with notch filters

[ppataki]

I was reading an article about room modes when an idea came to my mind
We can fix frequency-related room modes pretty easily with EQ (either using REW or more automated solutions like Dirac Live, Acourate, etc.)
But once the frequency domain is fixed there can still be temporal resonances that remain there and they can be really annoying especially in the <100Hz range

This is the Dirac Live optimized frequency response of my L+R speakers:

I still have a huge room-related dip that cannot be fixed but the rest is OK in my opinion

Now if I take a look at the Waterfall diagram (using 400ms time range) I can see that there are some frequencies ringing more than the others:

Some of these are pretty nasty......

So how to fix these in the digital domain?
I thought I would try some notch filters - those are very narrow filters to remove the signal between two frequencies; basically a surgical EQ

I was using T-Racks Equal to make it happen and I decided that instead of completely removing the signal with notch filters I would just attenuate it with traditional bell filters but using very narrow Q (20) based on the Waterfall results above (at 42, 50, 66 and 98Hz)

Here is how it looks like:

Now let's see what it did with the Waterfall:

I would say this is quite a difference!

It also has an impact on the frequency curve of course:

The biggest difference is visible in the T60M curves......

And actually when listening to it it sounds way better! There are no more audible long decays but the bass is still there, it is just not annoying anymore

I fine-tuned it a bit more by keeping only those filters where the resonance was audibly most prominent and also I have realized that even a reduction of 2-3dB is enough to get rid of the annoying ringing.

Let me know your experience once you tried it - I hope it will help others too

 

[boxerfan88]

I am with you on this.

After much trial and error, 9 mths back, I landed on a similar approach of using steep high-Q notch filters to manage room resonances.
The only difference is that mine is done inside foobar2000 using MathAudio's PEQ. (I don't have Dirac).

Before (43Hz, 55Hz, 70Hz rings past 300ms even up to 750ms)

Notch Filter Applied

After (far less ringing past 300ms) sounds really good to my ears.

 

[Shadrach]

Wish I had seen this thread before I started mine on a similar topic.
I've been experimenting with notch filters and shelf filters. I'm interested in what can be done with a loudspeaker system rather than headphones.
Sometimes there are room nodes that are difficult to deal with due to the penalty dB drop in the preamp. Being able to attnuate a single frequency would be very useful if the frequencies either side could be left untouched.

 

[Dillen]

6 months old thread, anyways.... just found it

Cool!

I have come to conclusion about the same, i use narrow high Q filters in my IOTA AVX 17 to begin with to get rid of the ringing peaks in frequencies in the bass region. After that i tame and shape the frequency curve manually to my liking within my Power amp (Inuke 6000dsp).

When i applying those high Q filters there's always a trial and error to get it right some times i can dampen alot and other times not.

To me it sounds best when the decay is short even if i have to sacrifice a slightly more uneven frequency curve (trying to have the best psycho acoustic curve).

 

[ozzy9832001]

I think this thread helps to illustrate that FR isn't always everything. Long decay can ruin the sound more than a dip or peak can.

 

[ppataki]

I think this thread helps to illustrate that FR isn't always everything. Long decay can ruin the sound more than a dip or peak can.

Absolutely! That was one of the reasons why I started it

See another post I made today about Dirac measurements - strangely (or not so strangely) the worse FR 9-point measurement sounds much better than the 1-point measurement with much better FR