Audio Editing in Premiere Pro: What They Don't Teach You

I used to be terrible at audio. Not just mediocre, actively bad. I would spend days obsessing over picture cuts, then spend maybe twenty minutes on the mix before export. Levels looked fine on the meters, so what else was there to do? Then I delivered a commercial where the voiceover dipped under the music every time the talent paused, and the client's audio engineer sent back a note saying the dynamic range was worse than a phone recording. That stung, but it sent me down a year-long rabbit hole, and I rebuilt my Premiere audio workflow from scratch.

What I learned: you do not need to be a re-recording mixer to produce professional audio. But you do need to know what Premiere offers, what to ignore, and which few techniques actually move the needle. This is everything I wish someone had handed me on day one.

Track Effects vs. Clip Effects: Know the Difference

This is the single most confusing thing about Premiere's audio architecture, and Adobe does not make it obvious. There are two completely separate places to apply audio effects, and they behave differently.

Clip effects are applied to a single clip. Select an audio clip in your timeline, open the Effects Controls panel (Shift+5), and apply something like a Parametric Equalizer. It affects only that clip, and it travels with the clip. Move it to another track and the effect follows. This is what you want for clip-specific problems: a de-esser on one interview subject who hisses on every S, or a high-pass filter on the one lav mic that picked up HVAC rumble.

Track effects live in the Audio Track Mixer (Window > Audio Track Mixer), which looks like a hardware mixing board. At the top of each track column there is a dropdown that says Default. Click it, choose an effect, and it applies to everything on that track. Put a compressor on Track 1 and every clip on Track 1 gets compressed. I always put dialogue on Track 1, music on Tracks 7 and 8, and sound effects on Tracks 5 and 6, so track-based EQ and compression on Track 1 means every piece of dialogue gets the same basic treatment without me adding effects clip by clip like a maniac.

The Three Effects That Matter (Ignore Everything Else)

Premiere's audio effects list is intimidating. There are dozens of options, most of which you will never touch. I use exactly three on almost every project.

1. Parametric Equalizer. EQ is frequency surgery, and most audio problems are frequency problems. On almost every dialogue track I make three moves:

• •High-pass filter at 80 to 100 Hz, Q of 0.71. Voices do not produce meaningful sound down there, but room rumble, air conditioning, and footfalls do.
• •Notch cut at 200 to 400 Hz, the mud range. Sweep a narrow cut (Q of 2 to 4) until the track opens up, usually around 250 to 300 Hz, and pull 2 to 4 dB.
• •Gentle boost of about +2 dB at 3 to 5 kHz, the presence range where intelligibility lives. Be subtle. Too much and your subject sounds harsh.

High-pass, mud cut, presence boost. Those three moves fix about 80 percent of dialogue problems. The rest needs more work, but start here every time.

2. Single-Band Compressor. Compression controls dynamic range. A voice that swings from whisper to shout is hard to listen to and impossible to mix against music. I usually set the threshold around -18 to -12 dB, a ratio of 3:1 to 4:1 (natural, not smashed), 10 ms attack, 100 to 150 ms release, and then use makeup gain to bring the level back up so it peaks around -6 dB. Last month I cut a six-minute corporate interview where the subject kept leaning into and away from the mic, swinging from -12 dB to -3 dB. A 4:1 compressor at a -15 dB threshold held it steady between -15 and -9 dB, and the mixer who got the files thanked me out loud in the session.

3. Hard Limiter. A limiter is a compressor with an infinite ratio. Nothing gets past the threshold, period. I put a Hard Limiter at the end of my dialogue chain with the ceiling at -6 dB as a safety net for laughs, chair squeaks, and sneezes. Chain order matters, and mine is always Parametric EQ, then Compressor, then Limiter. EQ shapes the tone, the compressor evens the dynamics, the limiter catches the peaks. Put the limiter first and you are limiting un-EQ'd audio while the compressor can still spike past it.

Essential Sound Panel: Friend or Foe?

The Essential Sound panel promised to make audio accessible, and it does, sort of. But it also trains you to accept mediocre results if you are not careful. I use it as a starting point, not a finish line. Select a dialogue clip, click Dialogue, and it applies auto-EQ, auto-Ducking, and auto-Loudness. The auto-EQ is decent for quick-turn projects. The auto-Ducking is where I get cautious, because the timing is often clunky, so I use it as a first pass and then hand-tune the generated keyframes.

The Repair section is the genuinely useful part: Reduce Noise, Reduce Rumble, DeHum, DeEss, and Reduce Reverb are iZotope-powered. I use Reduce Noise at 5 to 10 on almost every location recording and Reduce Reverb at 3 to 5 for slightly boxy rooms. Do not max them out, because they introduce artifacts fast. What I never trust as a final step is the Loudness auto-match. It gets you close to broadcast spec, but I have seen it overcorrect, so I use it to ballpark and then verify with the Loudness Radar.

Level Standards: The Numbers That Matter

This is where most editors wing it. Do not. Here are the actual numbers I work to, measured with Premiere's Loudness Radar (Window > Audio Meters > Loudness Radar). The integrated LUFS reading is what matters for delivery.

• •Web (YouTube, Vimeo, social): dialogue around -16 LUFS integrated, true peaks at -1 dBTP, music 10 to 14 dB below dialogue during speech.
• •Broadcast (TV, cable): -23 LUFS integrated (the CALM Act standard in the US), true peak -1 dBTP or lower, loudness range under about 8 LU.
• •Theatrical and festival: dialogue closer to -27 LUFS integrated, with a wider loudness range of 10 to 15 LU.

The Audio Handoff to a Real Mixer

Sometimes the budget exists for a proper mix, and when it does, your prep decides whether the mixer loves you or curses your name. First, clean your timeline. Every clip should start and end with short fades, no clicky edit points. Dialogue cuts usually need 3 to 6 frame fades, because hard cuts on dialogue sound like jump cuts for the ears.

Second, organize by track so the layout is predictable: primary dialogue on Track 1, secondary dialogue on Track 2, sync sound on Track 3, effects on Tracks 4 and 5, design on Track 6, and stereo music on Tracks 7 and 8. The mixer should be able to mute the music tracks and instantly hear a dialogue-and-effects version. This is the same track layout I build into my project template so I never set it up from scratch.

Third, export stems properly. This used to be the most tedious part of the handoff: soloing tracks one by one, setting in and out points, exporting, repeating. For a ten-track project that is at least ten exports, easy to screw up and easy to miss a track. Now I use Track Exporter, which handles the whole stem export in one shot. It writes each track as a separate file, properly named, same duration, sample-accurate sync. The mixer drops the folder into Pro Tools and everything lines up. I save about 45 minutes per project, and the one time I forgot to solo a track by hand (and sent a stem with music bleed) was the last time I did it manually. It lives with my other presets and plugins and for anything beyond a three-track mix it is not optional for me anymore.

What to Do About Bad Location Audio

Every editor has been there. The footage looks great, the performance is perfect, and the audio sounds like it was recorded in a parking garage during a windstorm. If the dialogue is salvageable, my triage is to high-pass aggressively (up to 120 to 150 Hz if there is rumble), apply Reduce Noise at 10 to 15 while listening for the underwater bubbling that means you pushed too far, add DeReverb at 5 to 8 for small-room slap, and use the Multiband Compressor to bring up the 1 to 4 kHz intelligibility range without crushing the whole signal.

If the dialogue is not salvageable, be honest with the producer immediately. Do not spend two hours trying to polish a turd. Say it plainly: this needs to be re-recorded or we will need ADR, and I can make it acceptable for a rough cut, but it is not deliverable. Clean it up enough that the story works, flag it, and let them decide if ADR is in the budget. Do not pretend bad audio is good audio. Clients remember.

Monitoring: The Thing Editors Skip

Your audio is only as good as what you hear it on. Laptop speakers lie, earbuds lie slightly less, cheap desktop speakers lie a lot. I edit on a pair of studio monitors on isolation pads, set in an equilateral triangle with my head so each speaker sits the same distance from me as from the other. Before I had monitors I used a pair of closed-back studio headphones that every sound person owns, and if you can only buy one thing, buy those. Check your mix on them, then on earbuds, then on laptop speakers. If it works on all three, it works.

Putting It All Together: My Starting Template

I have a Premiere project template I use for every new edit, with the audio tracks already named and Track 1 pre-loaded with a starting chain. Before I import a single clip it already has:

1. Parametric Equalizer: high-pass at 80 Hz, gentle notch at 250 Hz (-2 dB, Q of 3), shelf boost of +2 dB at 4 kHz.
2. Single-Band Compressor: 3:1 ratio, -18 dB threshold, 10 ms attack, 100 ms release, auto makeup gain.
3. Hard Limiter: -6 dB ceiling.

I adjust these for every project, but having them there means I never start from zero. The difference between no processing and basic processing is the difference between amateur and professional audio. Everything after this is refinement. Audio is not magic. It is a handful of concepts applied consistently: learn your EQ, learn your compressor, learn the difference between clip and track effects, and organize your timeline like a professional. The mixer who gets your stems, or the client who watches on their phone, will hear the difference.