Sound Advice: Match Your Levels Before the Show, Not During It

Every show has this moment. Walk-in music is sitting nicely under the crowd. The lights go down, the opening video rolls, and it hits the room 8 dB louder than anything before it. The A1 grabs the fader. Two cues later, a testimonial video plays so quietly the back half of the room is reading lips.

Nobody did anything wrong in the booth. The files just came from different places. One video was mixed by an agency for broadcast. One was cut on a laptop by someone in marketing. The walk-in playlist is streaming-service masters. Each one is at a different loudness, and the room hears every difference.

Peaks aren’t loudness

The usual check is to look at the meters and make sure nothing clips. That tells you the peaks. It doesn’t tell you how loud a file sounds. A heavily compressed promo and a quiet interview can peak at the same level and be 10 dB apart to the ear.

That’s what loudness measurement is for. LUFS (loudness units relative to full scale) measures the average perceived loudness of a whole file, the way people actually hear it. Two files at the same LUFS sound about the same volume, whatever their peaks are doing.

The fix: pick one number and normalize everything to it

Before the show, run every playback file through loudness normalization to the same target. That’s videos, stingers, walk-up music, and the walk-in playlist. Then set the playback channel once and leave it.

• Pick a target and stick to it. For event playback, somewhere around −20 to −23 LUFS integrated, with true peaks under −1 dBFS, is a sensible place to start (−23 is the European broadcast standard). That leaves headroom for the loud moments in the content. The exact number matters less than every file having the same one.
• Use the tools you already have. Premiere, Resolve, and Audition all have loudness normalization. Audacity has a free Loudness Normalization effect. If you’re comfortable in a terminal, ffmpeg’s loudnorm filter will batch a whole folder.
• Normalize copies, not originals. Keep the client’s files untouched, and label the show versions so nobody grabs the wrong one.
• Do it when files arrive, not at the console. Ask for content early and normalize it as it comes in. The video that shows up at 4 p.m. on show day gets the same treatment, and it only takes a minute.
• Still listen. Normalization gets you close. A video with a loud music bed and quiet dialogue might still need a nudge. That’s a small ride, not a scramble.

Same idea, smaller room

Podcasts have the same problem. The intro music, the guest on a laptop mic, and the ad read all show up at different levels. The listener fixes it with the volume button, usually right before they skip you. Apple recommends around −16 LUFS for podcasts. Mix the show to a target, and match the ad reads and inserts to the episode, not just to each other.

The short version

If you’re riding the fader for every video, the problem started before the show. Match the files once, set the channel once, and spend the show mixing the people instead of chasing the playback.