I always teach my colleagues, my students…heck, even my 2 and 5 year olds, that you should set the microphone input gains in a DSP mixer so that a standard talker averages 0dB. That scheme will give the system a great signal to noise ratio at the microphone inputs. It will give the echo canceller plenty of waveform to remove the reference from. It gives the site file author plenty of signal level to work with down the line. It’s the best way to set the gain on a microphone.
Until it isn’t.
My team and I were working in an extraordinarily reflective and reverberant space. All the surfaces in this fairly large conference room were hard: glass table, glass walls, sheetrock ceiling, concrete floors, large windows. It sounded terrible on the far end. Some user descriptions included “echo-y,” “like you’re in a cave,” “like you’re in a middle school gym,” “using a terrible speakerphone” and, my favorite, “like you’re in the belly of a giant space slug in the Hoth asteroid field.” (I really liked that guy. I felt like he got me. Anyway…)
Intelligibility was not optimal. There were plenty of early reflections hitting the microphones, which muddled the sound. Early reflections are the slightly delayed sound pressure waves that arrive after the pressure waves that came directly from my mouth. They bounce off the ceiling, the table, the walls, the windows…and since each wave takes a slightly longer path than the direct sound, they arrive at different times. This always happens in every space, but because there was so little absorption from the hard surfaces of the room, these early reflections still had a lot of signal. Usually, these hit acoustic tile, acoustic paneling on the wall, the carpet, etc., and die off quickly with little effect on the microphone audio. That was not the case here.
We played with the EQ a bit to roll off the low frequencies, but it made me sound tinny. Then, my colleague decided to set the microphones at -6dB in the mixer instead of 0dB. And, immediately, the audio was dramatically more intelligible. It sounded so much better that the client (that awesome Star Wars guy) stood up because he had never heard it sound that good. The room contributions (early reflections) were simply too strong. The microphones were set too high for the space. We made up the difference at the conference transmit signal, so the level was still strong going to the far end. The echo canceller in the DSP mixer was well done, so it didn’t skip a beat. We still had no echo and great duplex performance. Dropping the microphone level kept the direct sound present while making the early reflections from the room less of an issue. It was a great idea, and the client was very happy.
As my friends Pat and Brenda Brown say from Syn-Aud-Con (They rock! Check them out: synaudcon.com), “It depends.” Everything with audio depends on the particular circumstances. While that Set Mics to 0dB Rule works most of the time, there are certainly circumstances that require something different. For example, typically only Jedis should handle light sabers, but when your friend might die due to hypothermia, you KNOW you gotta light that puppy up to make him a Tauntaun sleeping bag.
When talking about audio, the surefire answer will always be it depends.