Here are 2 panoramas showing the areas we recorded in:

The living room has wood floors, little furniture, a couple rugs, a reasonably high ceiling, and feels quite “live.” Along the long direction, there’s a fireplace and staircase. Along the short direction, there’s shuttered windows and a wide walkway into the den. The den is relatively square, has a lower ceiling, and is capped on one end by a wall of sliding glass doors. In front of the den’s couch, across from its rug, along the wall that you can’t see much of in the top photo, is a large wall TV. Behind the couch, opens into a tiled, resonant kitchen. The den feels much less “live.”

We performed sound checks in these locations:

• LOC 1 – facing the fireplace, mic placed atop a throw rug that also contained our 3 chairs (see top panorama)
• LOC 2 – similar, but now facing the den (panorama 2)
• LOC 3 – in the den, on top of its rug, facing the TV

Below shows our M-S mic setup:

When performing the medley Kansas City Hornpipe / Bottom of the Punchbowl (KCH-PB) our orientation around the M-S mic is:

Preliminary listening led us to believe that the fiddle sounded better in the ensemble when it was a bit off center, and since fiddle and banjo frequencies can compete, fiddle was moved a bit closer to guitar (i.e. to the right of the dashed line). When performing Wild Hog (WH), because guitar sings lead and fiddle sings backup, we swapped positions. In WH audio clips, be sure to listen to the end, so you can hear the mix with instruments and voices, followed by just instruments. In hindsight, both vocals lean to one side, which is not ideal (it might’ve been better to have GIT sit on the other side of the dashed line). I am unsure what the best way to record our vocals is, especially since fiddle usually sings backup, and is not always harmonizing when guitar sings lead. Add some spot mics? Record vocals separately? Need to talk to Mario about this!!

In order to compare various sound tests, I “normalized” each M-S decoded stereo result via clip gain. For each test, M and S clip gains were identically increased, so that the M/S ratio remained constant. LUFS (short and integrated), True Peak, and K-20 peak were considered when setting clip gains. The stereo signals presented here are raw (no effects or EQ are added). For posterity, my normalization notes are:

After determining that fiddle should sit a bit off center, we preformed our first sound check in LOC 1. In TRY 1, the fiddle was about 1.5′ away from the mic, banjo about 2.5′, and guitar about 3′.

We thought the guitar should pop-out a bit more, so TRY 2 moves the guitar in .5′ and sounds like this:

We tried to use this latter configuration in all our location tests, but we were never as careful as we were when doing these first two trials.

It is worth mentioning that the “stereo width” heard in various locations differs because of the area’s geometry. M/S ratios were left fixed, so what I present is what was recorded. To reiterate: I did not “improve” any location’s result by modifying its M/S ratio. The following image indicates how much mid versus side mics contributed to an overall result for KCH-PB, where mid-mic audio is displayed on top, side-mic on bottom, and color indicates location (blue: LOC 1 TRY 1, purple: LOC 1 TRY 2, red: LOC 2, yellow: LOC 3).

Here’s how the locations sound:

My impression is that LOC 1 provides the most “live” and “in-the-room” sound. What do you think? When we record for real, I suggest we spend more time listening carefully to how each piece lies in the mix before recording for real. I need to talk with Mario about how to handle vocals. It sure would be nice to record them while we play our instruments but I suspect the complexity of getting a great mix goes up exponentially, not to mention it will be harder to pull off “near perfect” recordings. Your thoughts?