The feedback was at 2 kHz, clear as day. A howling, piercing ring that appeared every time the vocalist stepped within a meter of the monitor wedge. Standard procedure: identify the frequency, apply a narrow notch, move on. Simple.

Except Ricardo didn't apply a narrow notch. Ricardo applied a notch so narrow that light itself would struggle to pass through it. His parametric EQ showed -18 dB at exactly 2,000 Hz with a Q factor of 50. For those unfamiliar, a Q of 50 at 2 kHz creates a filter approximately 40 Hz wide — surgically removing one tiny slice of frequency spectrum while leaving everything else untouched.

The feedback stopped. Ricardo smiled. The vocalist opened her mouth and something deeply unsettling came out. Her voice had a hollow, metallic quality, like she was singing through a drainpipe in a parallel dimension. The missing 40 Hz slice happened to sit right in the middle of the formant region that makes a human voice sound human. She sounded like a text-to-speech engine from 2003.

'I sound like a robot,' she said, accurately. 'That's just the monitor mix,' Ricardo assured her. It was not just the monitor mix. It was every mix. The notch was on the main system output.

The Moral: Notch filters need appropriate width. Too narrow and you create audible artifacts; too wide and you affect wanted content. Use SonaVyx's AI diagnostic to identify the optimal Q and depth — the right notch kills feedback without killing the music.