Thursday night karaoke at O'Malley's Pub was a beloved local tradition. The rules were simple: sign up, pick a song, sing your heart out. The rules of acoustics were somewhat less forgiving.

The system was a pair of powered speakers on brackets and two wireless handhelds that had seen better decades. The first singer of the night — Karen, performing "Total Eclipse of the Heart" — held the mic six inches from the speaker. A howl erupted at 1 kHz. The bartender reached behind the bar and turned the gain down. Karen complained she couldn't hear herself. He turned it back up. The howl returned. This cycle repeated for every performer.

By the third singer, someone had discovered that cupping the mic made their voice louder and more resonant. It also turned the cardioid pickup pattern into an omnidirectional one, reducing gain-before-feedback by 6 dB. By singer seven, three patrons were holding mics directly against their mouths while standing in front of the speakers, and the bartender was developing a twitch.

The system had roughly 4 dB of gain-before-feedback margin when the mic was used properly. With cupping, proximity to speakers, and the cumulative effect of drinks on performers' spatial awareness, that margin went negative.

The Moral: Set gain-before-feedback limits during setup, not during the performance. SonaVyx Problem Detection identifies exactly where those limits are — so you can lock the fader and enjoy the show.