Best Practices in Q-SYS Gain Structure (Part 2)

Welcome back! Let's pick up where we left off.

Reaching nominal gain in Q-SYS is very simple:

Plug in your signal and adjust the preamp gain until the input meter level reads about -20.

That’s it!

You are going to get a far better result if you calibrate every preamp

so that you have a uniform -20 dBu nominal,

rather than having to compensate later in the signal chain in cross points and matrix mixers.

This is true for every input to the system:

Microphones, Program material, conferencing feeds….everything.

There’s one major additional concern when it comes to microphones.

Microphones will have their own noise floor based on the room they're in,

the amount of noise in the room, the quality of the microphone etc.

According to the Avixa standard for mic speech intelligibility, the minimum signal to noise ratio is 15dB.

If the ambient noise level at the microphone and your nominal level of the talker is less than 15 dB

then you’re going to find you're in trouble.

Unfortunately, you can find yourself in that situation all too often,

particularly when using common ceiling microphones next to air handlers.

Unfortunately, no matter how great your DSP might be,

there's no way to fix the signal to noise ratio of a given input.

AEC algorithms do provide some steady-state noise reduction,

but when you have poor signal to noise ratio,

it’s not going to be able to distinguish between signal and noise.

That’s why the most crucial part of the signal chain is the integrity of that first sample.

You can gauge that microphone intelligibility in Q-SYS at the meter.

If nominal level of -20 dBFS has been achieved,

but the noise floor is above -35 dBFS, we cannot ensure intelligibility.

Calibrated to -20dBFS nominal, if the noise floor at the input meter is BELOW -35dBFS,

then it should pass the intelligibility test.

Once we sample the mic signal,

most integrators will apply some microphone processing to get the best overall performance.

The most common mic processing combination seen in the field is high-pass filter,

parametric EQ and low-pass filter.

The high pass filter is used to remove low-frequency noise, typically set to 60 or 80 hertz.

Parametric EQ is meant to remove frequencies that might cause feedback

and equalize the natural response of the mic in the space.

The low pass filter is used to remove any unwanted high-frequency noise,

and each will vary based on application.

For vocal performance, you’d like to keep as much high frequency as possible,

while that can be sacrificed in conferencing applications.

Generally speaking, the goal should be to maintain as much high end frequency as you can,

however the SIP codecs in a conferencing application compress most of it anyways,

so you can choose to remove more of the high end energy for the sake of clarity.

Keep in mind that different components of speech have different frequency ranges.

Vowels and the fundamental human voice are relatively low frequency.

Consonant sounds are in the middle range, while sibilance sounds are in the higher ranges.

If you are looking to get a little more intelligibility,

look to manipulate the constants and the sibilance.

If you are encountering feedback, you can address it using an RTA in the space or just by ear,

but the Notch Feedback Controller component

is a perfect tool for figuring out exactly what frequencies you're feeding back.

You can put this into your signal path and follow these instructions to force the system into feedback.

First, set a fixed filter count to 0 and set a dynamic filter count to 5.

Then set the “Min Dynamic Frequency” to 80 Hz and the “Max Dynamic Frequency” to 20 kHz.

Finally, find the threshold of feedback and go slightly over.

The resulting feedback frequencies will be easy to identify in the NFC block.

In this example, you see feedback in two specific ranges.

After you are finished, we do not generally recommend leaving these in the signal path,

as you’d typically reserve these for use in very difficult environments.

It’s more effective to use this as an analysis tool and then create a parametric EQ that is a bit more gentle

to help mitigate those frequencies without affecting too much of the overall tone.

Keep in mind that this will be an iterative process.

For example, you should always start with the lower frequencies

because those are typically going to be fundamental of the problem frequencies.

If you fix those, you're less likely to have feedback in the higher frequencies as a result.

When you have transferred the information that you've learned from the Notch Feedback Controller

over to your parametric EQ you should have mitigated most of the risk of feedback.

Alright, that's a good place to pause. Move on to the next video whenever you're ready.