I recently took a crack at improving the sound proofing between the main floor of my house and the basement using acoustic treatments. As evidenced by a quick Google search, about a million people have done this before. The idea is simple: construct a rigid wooden frame to house some type of absorptive material (such as stone wool insulation), and cover that in a porous fabric (such as burlap) to hold it all in. The effectiveness of such a contraption (we’ll call it the panel) is frequency dependent – thicker panels are needed to block lower frequencies, which have longer wavelengths and can penetrate through and around objects more easily than high frequencies.
It seemed as though the majority of sound leakage was happening via our stairwell and basement door, so this is where I concentrated my effort. I used 1” x 6” pine boards to construct the frame, Rockwool Safe’n’Sound insulation, and burlap fabric. The material cost was about $120 CAD here in Newfoundland (fun fact: the Rockwool I bought is about $30 cheaper in Ontario than it is here owing to freight costs). A pretty minimal set of tools is needed for the build: a mitre saw to cut the frame pieces and supports, a cordless drill to fasten them together, and a stapler to attach the burlap. I borrowed most of my tools from the St. John’s Tool Library.
I fastened the edges of the frame using wood screws and L-shaped brackets. I added some extra angle supports to each corner for additional strength. I used two layers of Rockwool battens in order to fill the 6” deep frame. Here’s a picture of the panel mid-construction:
And here’s a picture of the finished panel fitted in front of the basement door:
I decided to evaluate the effectiveness of the panel by measuring the frequency response between the upstairs of my house and the basement both with and without the panel in place. I decided to use Farina’s log sine-sweep method, which is particularly affective at rejecting loudspeaker distortion introduced during the measurement phase. Given that I needed to drive my loudspeakers quite hard to ensure adequate sound propagation through to the basement, I expected distortion would be an inevitable by-product in my measurements.
After making recordings of the sine-sweep both with and without the panel in place, I deconvolved the measurements using the inverse sine-sweep, and removed the distortion by-products which appear at the start of the response (see Farina’s paper for more on this). I performed the deconvolution using fast partitioned convolution (you can read more about convolution in a previous blog post). Below is a plot of the 1⁄3 octave smoothed frequency response I computed from these measurements:
As can be read off the graph, the panel is effective at frequencies of 200 Hz and higher (note the x-axis is on a log frequency scale). At the peak frequency (approximately 1000 Hz), there is a 16dB reduction in the acoustic energy transfered between floors. That seems pretty good to me. I also made a few informal tests, for example, I recorded myself singing and playing guitar. Without the panel in place the sound transfer was clear enough that I could make out what lyrics were being sung; with the panel in place, my playing was just barely audible. Overall, I’m quite happy with the performance of the panel, and would consider building more (e.g., for acoustically treating a studio space).