My wife and I recently bought a house (well, she bought a house and she’s letting me rent a room) and for the first time in my life I have the opportunity to do whatever I want with my office. I installed a few different sound absorbers, taking measurements along the way, and here are my results.

The Room

The first time I walked in I knew it would need some help. When I clapped my hands I heard something ringing and it was more difficult than it should have been to understand someone talking a few feet away.

Let’s estimate where the room starts to transition from modal behavior to diffusion by dividing the three times the speed of sound by the room’s smallest dimension (3c/RSD).

3390 / 9.02 = 376Hz

Let’s look at the Room Simulation in REW.

Since the length and width are almost the same we can see two axial modes lining up nicely around 50, 100, and 150Hz. I hope some absorption will help with that.

I took four sets of measurements:

1. Empty
2. With ceiling mounted absorption panels
3. With wall mounted Audimute sheets (and the ceiling panels)
4. With furniture (and ceiling panels and Audimute)

Let’s get a global view of the overall changes with a spectrogram.

And now the RT60 over 1/3 octave frequency bands.

I’m looking at the value in REW called Topt RT60 because:

The start point for the classical T20 and T30 measures of RT60 is where the Schroeder curve has dropped 5 dB below its peak. That works well in the large spaces for which RT60 is most applicable, particularly if the source used for the measurement is omnidirectional. In domestically sized rooms using normal, directional loudspeakers as sources the initial drop of the Schroeder curve is quite sharp (the Early Decay Time is quite short) meaning the -5 dB point lies within the early decay region rather than the diffuse field region. That in turn means the T20 and T30 figures underestimate the RT60 time. Where the EDT is much shorter than the T30 RT60 figure REW’s Topt RT60 calculation uses a start point based on the intersection of the EDT and T30 regression lines, to determine a point that lies within the diffuse field region. It then tests each possible end point in 1 dB steps and picks the one that gives a regression line with the best linear fit. That produces a more reliable RT60 figure.

REW user manual

I think it’s pretty clear that the biggest improvement here is around 400Hz. There’s a bunch of energy in the midrange that just made it nasty to try to have a conversation in here. And it’s no surprise that the ceiling absorbers made the biggest difference. They were also the biggest pain to install, but totally worth it. Thanks to Matt Bombich here in Minneapolis for those.

Also keep in mind that the measurements are not made with each material installed independently. I didn’t take down the Rockwool ceiling panels so I could measure the Audimute sheets by themselves. The Audimute measurement shows the improvement on top of the ceiling panels and then the furniture on top of all of that.

Those axial modes I hoped to tackle did not see such a significant improvement. I’m not an expert on absorbers, but I assume I would need some kind of low-frequency monsters.

Did you notice the 60Hz hum in every measurement except for the final one? I’m not sure what finally killed it.

Empty

You can see the room modes predicted around 50, 100, and 150Hz.

With 3 Ceiling Absorbers

With 2 Audimute Panels

The Audimute panels gave a nice improvement to the low-mids.

After I purchased them I discovered that they are also available in green. Instead of painting the wall I could have simply hung green panels and covered them up with the black curtain when they weren’t being used. 🤷🏻

With All Furniture

The only thing you can’t see in this photo is a twin mattress I put on top of that book shelf for even more absorption.

Overall, I’m really happy with the way the office sounds. The lighting is a problem, but I can figure out a better layout later.

Here’s a link to all of the impulse responses I recorded. Maybe you can do your own analysis and find some things I missed.