Room modes can make your mix sound flabby and are most prevalent in small rooms. A few properly placed notch filters can help and in this article I am going to show you how to measure your room and place the filters using Smaart.
- Precisely placed notch filters can tighten up your mix, but it’s easy to overdo it. Listen and audition.
- Smaart will not average multiple IR measurements. You’ll need to do that in another app.
In general, the process is simple. The audibility of room modes is determined by duration so all you need to do is observe a Spectrograph or Waterfall of your room’s impulse response. The tricky part is getting the right impulse response.
The quality of the impulse response is important because notch filters are very narrow. You need accuracy. The more measurements you take, the higher the accuracy.
Once you take all of the measurements, you could simply look at them one at a time and attempt to find the trends among them, but a faster way is to create an average. Unfortunately, Smaart doesn’t have this functionality. Fortunately, I have a workaround for you.
First, we need data. Let’s measure the impulse response.
Measure the impulse responses in Smaart
Without too much explanation for why I have chosen these settings, here’s what I recommend for the impulse module in Smaart.
- FFT: 128k
- Averages: 2
- Signal: Pink sweep triggered by IR
- Level: 20dB above the noise floor from 20Hz to critical frequency* (Don’t stress about this. If you were already taking transfer function measurements and getting actionable data through the LF, then your sig gen level is probably fine.)
- Place mic at head height anywhere in the audience. Room modes are not distance dependent.
- Press play.
- File > Save impulse response.
- Repeat six times at six random locations. If the audience and room are symmetrical, you only need to measure half of it.
This should be one of the final steps of your system tuning work. You’ll want the entire system ready to go since you are measuring its interaction with the room.
Here’s the sound system I’ll use for this article.
Here are the mics. They look closely spaced because the audience was only that deep.
Here’s what the first measurement looked like.
Now let’s create the average.
Workaround #1 – Easy/Expensive
The easiest, yet most expensive ($1,400) workaround would be to buy FIR Capture. Pat Brown used it to teach me this process at his OptEQ seminar during InfoCOMM 2019. Let’s go through it.
- File > Import first impulse response (IR). Repeat for all IRs. No time window necessary.
- Normalize all at 100Hz.
- Create power average. (Or optionally, right click on the IR plot and choose Sum Multiple IR to normalize arrival times and preserve the IR)
- Observe waterfall plot. If necessary, adjust time window for better resolution.
Here’s all of the IRs imported.
Here’s the average.
And here’s the waterfall plot where I have identified two room modes.
Workaround #2 – Medium difficulty/cheaper
If you don’t want to make the financial investment and the time to learn a new piece of software right now, I have an alternative for you: Averager.
This is an app from Eclipse Audio who you might know if you use FIR Creator. For $50, it’s a nice utility to get this job done.
- Preferences > Transform size > 131,072 (maximum)
- Load > Select directory with your IRs.
- Uncheck any files you don’t want to use.
- Optional: Select the IR with the greatest distance from the source as the reference. Delay > Time Align all to Reference.
- Gain > Normalize to a frequency range of 90-110Hz.
- Average > Averaging mode > Power
- Save > File > Format > WAV
- Set IR end to maximum (1,365ms).
Back in Smaart…
- File > Load impulse response and choose the file you just saved.
- Calculate Spectrograph at 16kHz with 99% overlap.
- Adjust upper and lower thresholds to discover room modes.
This method is more challenging than the previous because the graph is harder to read.
Workaround #3 – Medium difficulty/cheapest
Room EQ Wizard is a free app with some great functionality. Although it can create a power average, it will not generate a waterfall or spectrogram of the average. Your two options are to generate a vector average instead or open the average you created in Averager. You’ll get slightly different results, but essentially the same frequency information. It looks like this.
I still call this one medium difficulty because of the hoops you have to jump through.
Treat the room modes with EQ filters
Now that you have identified the room modes, treat them with narrow band (Q > 10, BW < 1/6oct) filters and listen to the results.
Here’s what my filters looked like in Vu-Net.
As you can see, I decided to audition a bunch of different filters.
Here’s the average IR of my room post EQ in FIR Capture.
And here it is in Smaart.
And here it is in REW.
WARNING: YOUR RESULTS MAY VARY
While listening in the audience, I auditioned each filter and discovered how easy it was to overdo it. A little help from the notch filter tightened up the mix, but too much and it lost its life and excitement.
What’s a room mode?
Resonance: wavelengths that “agree” with a volume.
a pressure wave that decays more slowly than those of the surrounding frequenciesdaytonaudio.com
Standing wave: non-propagating, it’s “standing” in space because it’s reflecting back and forth between two surfaces or nodes.
In physics, a standing wave, also known as a stationary wave, is a wave which oscillates in time but whose peak amplitude profile does not move in space. The peak amplitude of the wave oscillations at any point in space is constant with time, and the oscillations at different points throughout the wave are in phase.Wikipedia
Room mode: now we take duration into account. If the standing wave is standing around for longer than its neighbors, it might be a room mode.
Room modes are the collection of resonances that exist in a room when the room is excited by an acoustic source such as a loudspeaker.Wikipedia
*What’s critical frequency?
Critical frequency is a milestone in the transition of room acoustics from lower density modal behavior to higher density geometric behavior. It can be estimated with by dividing 3,390 by the room’s smallest dimension (3c / RSD).
If you are working in arenas all the time, you’ll never need to worry about it, but if you are working in small rooms it can give you some insight into the behavior of your room.
Also, the first mode can be found dividing the speed of sound by twice the room’s longest dimension (c/2L).
Questions I didn’t answer
Wouldn’t the ceiling need to be 25ft away or shorter to have a room mode at 132Hz?
Yes, because 3,390/25=137Hz. At the time, I neglected to consider this. I never measured the ceiling, but it was probably closer to 30 or 40ft, which would put the critical frequency at about 113 or 85Hz.
Why was I seeing resonances above the critical frequency?
My only idea is that critical frequency is a milestone for a transition not a true/false verification.
What do you think? (read Michael’s response below)
Also, have you tried measuring and treating room modes? What were your results?
Smaart® and the Smaart logo are registered trademarks of Rational Acoustics LLC and are not affiliated with Nathan Lively or Sound Design Live.
Michael Lawrence says
Hey Nathan –
Your notion of critical frequency (commonly called Shroeder Frequency) being more of a milestone is right on the money. The idea is really well presented in the Master Handbook of Acoutics and also in Floyd Toole’s “Sound Reproduction.” If you’ve got a copy of Sound Systems Engineering, check out the graphic on page 215 (fig 12-1).
It’s sort of a rough indicator of when we can expect the wave behavior to transition from modal to specular. It’s so rough in fact that Schroeder himself changed the constant in the formula by an entire octave between his original paper in 1956 and his updated one in 1996. So don’t be surprised if you see modal behavior popping up above Schroeder.
Also, to explain the 132 Hz mode you found in a room that’s seemingly too big, you can get higher order modes just like you can get harmonics on a string. So if 66 Hz fits between a room’s boundaries as a first order mode, you can expect some action at 132 as well. You can also see 132 without seeing 66, as the pressure maxima will be in different spots. For example, the center of the room for a first order horizontal axial mode is going to be a pressure minimum, but a maximum at second order. And so on. Depends where you sit, which is one of the reasons why you have to be careful with EQ.
Room EQ Wizard has a great room mode simulator, and you can see the higher-order modes created if you disable all modes rather than a single axial mode. This is part of the mechanism behind Schroeder – all the room modes (axial, tangential, and oblique) have higher orders, and since our frequency spectrum is logarithmic, the linearly spaced higher order modes start to stack up with more and more density in higher octaves (not unlike how a a single DFT gives more bin outputs in a higher octave). Once they’re sufficiently dense, it just becomes…reverb.
Nathan Lively says
Thanks for the clarification Michael!
BTW, have you tried it? What were your results?
fedele de marco says
wel defined, spaced, isolated, etc….”Room Modes” in the Very Very…very, large Room in the images? Simply you are measuring something else, not room modes. Take a look at what you have aroud and you will found the answer.
I also read in what Michael wrote “once they’re sufficiently dense, it just becomes… reverb.”
No… the “once they’re sufficiently dense” is something related to our “Tonal perception”, Reverb is another thing….reverb is something like “sound persistance”, something related both to TIME and Tonal perception.
Very good explanation
Nathan Lively says
Thanks Fredric! Have you tried it?
Bodo Felusch says
please allow me a little rant about it 🙂
you lost me at:
„1. Place mic at head height anywhere in the audience. Room modes are not distance dependent.“
Hell, they are! …that`s why they exist
you may have mistaken this with reverberation frequency response in large rooms/venues?!
Please also think about all the homestudio guys watching your videos. they will fuck of their mixes completely! never notch room modes because you kill notes.
Let`s say you place 41/65/145/330Hz …is that something you would do? …then have fun with the next song in C major.
Also please point out clear, that you can not eleminate room modes with EQ´s!
They are physical room acoustics artefacts that need physical treatment!
An EQ can only “adjust” the ammount of magnitude error at “one” place. The time domain (impulse response smearing) will stay the same. …and this is your real problem. Let´s say you have a certain frequency that dominates in the whole venue by 10dB, why should I choose a notch filter over a parametric? You could say “to eleminate the room mode” But this is wrong you can never eleminate a room mode by EQ, but you can avoid the excitement of it with a notch filter, but hey you pay with a note …lets´s say 55Hz …and tell the band not to play a song in B.
sorry for the rant, mate.
I disagree here!
Nathan Lively says
Thanks Bodo! I appreciate your dissent and calls for clarity. That’s why I do this. 🙂
What I meant to say is that your distance from the source doesn’t matter, but yes, the distance of the walls and architectural features is what creates them. Thanks for catching that.
Let’s say you’ve measured your mains and identified a problem resonance in the 50 – 100 hz range, would you apply a small notch to your subs as well as your mains?
Nathan Lively says
Hey James, yes, that’s the idea. Identical filters.
Michael Curtis says
Great stuff there. Love how you have the three approaches to getting great data.
Addressing the room modes takes care of the anti-nodes of the standing waves. Do you ever address the nodes (cancellations) of others? Or feel like taking care of something that’s both an antinode and taking a long time to decay?
Michael Curtis says
EDIT: Or do you feel like taking care of something that’s both an antinode and taking a long time to decay is a bigger win?
Nathan Lively says
Hi Michael! EQ can’t fix a cancellation. It can help reduce areas of build up. Is that what you meant?