My Home Studio - Construction and Design

Adding Room Treatment

The next step was to start adding some sound treatment. If you read the various newsgroups dedicated to acoustics, the most common word by far is "703", which refers to Owens Corning model 703 fiberglass. One might get the idea from reading these groups that this stuff is some magical substance, and indeed it was very hard to find. You won't find it in your local hardware, and even finding a way to order it was difficult. But it's essentially fiberglass insulation just like you probably have seen using in between studs in houses, except that it's very dense, so you can get the effect of a great deal of fiberglass in a 1" thick panel. It's also intended for acoustic uses, so absorption information is available for it.

I ordered 24 2X4 panels of 703, and filled most of the ceiling with 703. I just put the panels between 2x4 studs on the ceiling, holding the panels in place using screws with picture hanging wire run between them, as seen here:

I also placed 703 around the room in various places, temporarily, to experiment with how much sound absorbing material I might need and where. I plan to put some nice looking acoustic foam around the mix location, to create a reflection free zone, so I put up some 703 there for now. I then put what was left in the corners and around the walls of the room. I also had 4 Real Traps that I had ordered originally to place in different room, so this was my first chance to experiment with those.

I basically proceeded in 3 phases, 1) the 703 on the ceiling and around the mix position (14 2x4 panels of 703), 2) adding the rest of the 703 (24 panels altogether) scattered around the room, and 3) placing the four Real Traps around the room in addition to all the 703. I made test recordings for each of the setups and also used ETF again to measure the results.


Cutting to the chase, how does it sound? There is certainly a dramatic difference between the recording in the bare room and a recording with even just a small amount of 703 deployed:

Experiment 1: 703 on the ceiling and mix position (14 panels or 112 square feet)

Sample 3: Neuman TLM103 18 inches away

Sample 4: Berhinger ECM8000, 7 feet away

The recording is still a bit distance, with somewhat of a ringing sound, but it's better.

Next, I scattered the rest of the 703 around the room and re-recorded:

Experiment 2: 703 on the ceiling and spread around the room (192 square feet)

Sample 5, Neuman TLM103 18 inches away

Sample 6: Berhinger ECM8000, 7 feet away

The change is more subtle, but the sound seems a bit more direct to me. Next, I brought in the 4 Real Traps I had purchased. I simply sat them in the room, 2 in corners, and 2 against walls. This is not a proper installation, so I'm not expecting to get the full impact. Again, the difference is subtle:

Experiment 3: 703 on ceiling and spread around the room + 4 Real Traps

Sample 7, Neuman TLM103 18 inches away

Sample 8: Berhinger ECM8000, 7 feet away

Finally, I decided to try a closer mic'd recording. My goal is to have the room sound good enough to be able to mic a bit further away, but most of my previous recording have been done, out of necessity, from about 8 inches, so it was interesting to see how close I was to having this room be good enough for close micing. The result, below, seems reasonably acceptable (I didn't spend much time tweaking mic positions and so on)

Experiment 4: 703 + Real Traps, close mic'd

Sample 9, 2 Neuman TLM103s, in ORTF configuration, 8 inches away

ETF Analysis

Before I made each recording, I used ETF to collect data for each configuration. Let's see what the measurements tell us, and see if the data matches what the recording seem to indicate. First, below is the basic impulse response with only the initial 703 deployed. There is still a sharp reflection at about 8 ms, but most of the other spikes that appeared in the bare room are gone, or at least considerably tamed.

Impulse Response of initial 703 on ceiling

ETF can be used to compare multiple measurements, superimposing the graphs. Here is the original impulse response of the bare room (in black) superimposed on the above graph (in red). I think the reason the first large reflection doesn't line up is that I did not have the measurement mic in exactly the same location in ach measurement. In spite of this, it is fairly easy to see some improvement.

Comparison of bare room vs initial 703

One interesting graph produced by ETF is the Energy Time Curve (ETC), which shows the energy in the room from the impulse response over time. Overlaying the four measurements I have collected  so far, we can compare the different amounts of absorption in the room. The black line below is the bare untreated room, red is with 703 on the ceiling and mix position, blue is 703 scattered around the room, and green is with the additions of the 4 Real Traps. As one would expect, with each added absorptive element, the energy in the room declines faster.

Energy Time Curve for 4 different room treatments

I also did a rough calculation of the expected reverb times (RT60). Ignoring the rest of the room elements, and computing the RT60 for the 24 panels of 703 + the 4 Real Traps gives me the chart below. The "sq ft" column indicates how many square feet of each material are present. The "A" columns give the absorption coefficient for that material at each frequency. The "SA" colum is the number of sabines of absorption, and is simply the A value times the square footage. The sabine are summed up for each frequency, and then the RT60 value is computed as 0.049 * Room Volume/sabines, in the last row.  So, with all elements in place, one would expect about 1 second RT60 at 125 Hz, 0.93 at 250 Hz, down to 0.43 at higher frequencies. Real numbers would likely be lower because there are other absorptive elements in the room (drywall, etc), but this gives me a ballpark number.

Calculation of expected RT60 with full treatment

So,created in Excel. The black line is the bare room, while the Green is the fully treated room. I'm not sure why the times are so long at very low and very high frequencies, perhaps this is an artifact of the measurement. Times in the mid-bands seem to match expectations. At 1KHz, the RT60 has been reduced from around 700 ms down to under 400 by the ceiling 703, and reduced further as more 703 was deployed.

RT60 Comparison with different treatments (Milliseconds vs Frequency)

I graphed both the low and high frequency response with the various treatments, but the differences are sufficiently subtle that it's not clear to me if the changes are due to the treatment, or simply rand

The plot for the fully treated room is marked different. The gullies do not seem to be quite a sharp and extreme, and the response of later time slices shows the reduced reverberation time

Waterfall plot of low frequency response for the fully treated room

What's Next

All in all, this step seems to be starting to look good, installing hard wood floors, adding trim, and so on

I seem to still hear a bit more hard reflection in the recordings than I'd like, so even though I don't want to reduce the overall RT60 much further (except perhaps at the front where I'd place mics when recording.