Checklist Item Under Test: 6.31: Confirm that the speech reinforcement system is stable (no feedback) for the entire talker and listener areas specified.
Reasoning: Mix-minus systems can add a level of intelligibility to large rooms, using properly mixed microphones and loudspeakers to intelligently distribute the audio. The basic idea is to only send microphone audio to loudspeaker zones that are far enough away from the talker to require reinforcement, and only do so with enough level to enhance the direct sound from the talker.
Each loudspeaker zone gets a unique mix of microphones, minus those closest to those loudspeakers (to keep the system stable); hence, the term “mix-minus.” It can work great when properly deployed, allowing for natural-sounding conversations from across large rooms. Ceiling microphones, however, should not be included in properly deployed mix-minus systems.
The Story: I understand the allure of ceiling microphones. They can provide more flexibility for rooms that require several different table configurations. They can’t get covered by papers (without someone really putting some effort in). Coffee can’t spill on them (again…without someone really making an effort). If the CEO really wants that $150,000 Austrian glass and Italian marble boardroom table where people get fired for even thinking about drilling holes into it, ceiling microphones offer a viable solution. They never sound as good as getting the microphone closer to the talker’s mouth, but they work for conference-only (send) microphones. However, I must draw the line at attempting to use ceiling microphones for locally reinforced mix-minus systems.
From a quality assurance point of view, the intent to use ceiling microphones for mix-minus should be nixed at the design review. It should definitely raise red flags if it is even part of a DSP site file at a staging. However, for the sake of argument, let’s just say it actually makes it all the way into the field. I know what you’re thinking: “What’s the big deal? It’s a microphone. I have loudspeaker zones up the wazoo. I’ll be fine.” My friend, no…you will not be fine.
I could take you the mathematical route and get all into PAG-NAG calculations, and how the NAG (Needed Acoustic Gain) will most likely exceed the PAG (Potential Acoustic Gain). If the gain we need is more than the gain that the system can geometrically, potentially provide, it is a bad thing. When NAG > PAG, it proves that the system will be unstable. But that would be boring. Let’s just think about it empirically.
Our empirical system will have two zones, each with a user, a ceiling microphone and a loudspeaker. The purpose of the system will be to allow User 1 to converse across the room with User 2 without straining their voices. Our mix-minus is set up such that audio from Ceiling Mic 1 only goes to Loudspeaker 2, and audio from Ceiling Mic 2 only goes to Loudspeaker 1. The idea being that User 1 talks, is picked up by Mic 1 and the audio is sent over to User 2 via Loudspeaker 2. The reverse is also true for User 2 (but replace all the 1s with 2s). Simple enough, right?
However, the system will not function. It goes into feedback every time the levels go above 54dB-SPL (slightly below conversational levels at 1m), measured at the user’s ear level. How does this happen?
We know that sound pressure levels (SPL) drop off by 6dB as they travel, every time the distance is doubled. If a loudspeaker is producing 76dB-SPL at 1m, when listened to at 2m, it will measure 70dB-SPL, then at 4m, it will measure 64dB-SPL and so on.
Let’s say that our loudspeaker is six feet away from the user. The room has 10-foot ceilings and the user is seated at a table. However, because the loudspeaker and microphone are above the user, the loudspeaker is only three feet away from the ceiling microphone. That means audio from the loudspeaker will always be 6dB louder at the ceiling microphone than at the user’s ears. Thus, if the user is listening to audio at 54dB-SPL, that same audio will measure 60dB-SPL at the ceiling microphone. It will be louder at the ceiling microphone than at the ears of the user.
This also means that, if the user wants to use the system at conversational levels (listen and talk at 60dB-SPL at 1m), not only will User 1’s speech be 54dB-SPL at Mic 1 (six feet away, talking at 60dB-SPL at 1m), but the loudspeaker audio will be 66dB-SPL at the same microphone (only three feet away from the mic, but throwing six feet to be listened to at 60dB-SPL). The loudspeaker audio is 12dB hotter at the microphone than poor User 1’s speech. Yikes!
Now, let’s take a look at what’s happening in our mix-minus system. User 1 talks into Ceiling Mic 1. It gets sent to Loudspeaker 2. At this point, User 2 hears the audio, but Ceiling Mic 2 hears the audio louder than the user. So, Ceiling Mic 2 takes that same audio (an elevated User 1 signal), and sends it back to Loudspeaker 1 (as per the mix-minus). And then, guess what happens? The original User 1 signal (which was picked up by Mic 1, amplified by Loudspeaker 2, picked up by Mic 2) gets amplified again, and then sent back to Loudspeaker 1. The entire cycle starts to repeat itself over and over, but with an amplified signal after each iteration. The result is that all-too-familiar squeal of an unstable system. The audio loops from mic to loudspeaker to other mic to other loudspeaker, like a figure eight.
There are only two solutions. The first is to bring the microphones closer to the mouths so the dominant audio being captured is from a talker rather than a loudspeaker. The second solution, unfortunately, is to simply kill the mix-minus functionality of the system. It just will not work. Ever.
Ceiling mics and mix-minus just don’t mix.