When it comes to echo cancellation, microphone gain structure is essential to your success.
Let’s face it. When you break down today’s digital signal processors for audio and video conferencing, all you really have is a metal box with a little green guy inside running faders up and down. Although you tell him things like why, when, and how fast to perform his duties, he’s still relying on you 100% to accurately set up your microphones, otherwise known as the ‘front end’.
After working in this industry for over 16 years and programing many echo cancelled DSP mixers, I’ve seen that nine out of ten issues are due to the same cause. Our little green friend (let’s call him Simon) is usually sampling incorrect information due to the gain structure not being set properly on the microphone input. When this happens, you’ll see what could be referred to as a “domino effect.” Although Simon won’t admit it, our friend is nothing more than a piece of electronics that can, and will, make mistakes. However, when he is receiving the correct information, the number of mistakes he makes is reduced dramatically.
Think of it as if you were an individual who needs to wear glasses to see properly. You drive down the road wearing your glasses, and everything seems fine. Sure, you could have an accident at any time; but the odds are typically slim. Now pull over, take off the corrective lenses you love wearing so much, head on down that open highway, and see what happens. Just as we aim for 20/20 vision with our eyes, in our industry, we also aim for a standard volume level that the average talker uses when speaking.
Believe it or not, the average person speaks right around the 64db mark at a one-meter distance. Now, this isn’t like your average parent, at home with three misbehaving little ones, trying to deal with them after a long day of work. This speaking level is more like the average person talking normally during a casual conversation. Understanding this level will help you tremendously with your setup.
Here are a few tips and procedures to follow to keep in mind while setting up your microphones for conferencing.
Have the proper tools:
- A Sound Pressure Level Meter should be handy.
- A device that will serve as a signal generator will be a huge help.
- Understanding the software of the DSP to which you are connected. I can’t emphasize manufacture training enough. Most manufacturers will be happy to provide training to you, at little or no cost.
Although most people will simply use speech as their source, and watch the meters within the applied software package, I’ll briefly run through another method that is arguably more accurate.
We mentioned earlier that the average talker speaks at 64db at a one-meter distance, and we know that in almost every echo-cancelled DSP we want to see 0db from the signal being picked up by the microphone and reflected on the input. Using our signal generator, we can measure one meter away from the microphone element and place the generator there. Turn the signal on and set it for pink noise. We use pink noise for this process, as pink noise has equal energy per octave. Using your SPL meter, adjust the level of the noise to be 64db at the microphone. Now we are simulating speech, and can set our input levels accordingly to peak at 0db.
Regardless of what DSP you choose for your echo cancellation and mixing, setting your inputs peak at 0db using the method above will ensure Simon has the correct information to sample. When Simon is happy, everything works perfectly.
Today’s digital signal processors for audio and video are so advanced that I ask the question, in your experience regardless of most other settings, when your inputs are set properly, doesn’t your audio system sound pretty good?