Loading tutorials…
Loading tutorials…
Bad audio kills a video faster than bad video. This walks the full Premiere audio workflow: cleanup, EQ, compression, leveling, music ducking, and the loudness targets that distribution platforms require.
Who this is forVideo editors finishing videos that need to sound as good as they look. If your dialog is uneven, your music is overpowering, or platforms re-normalize your audio after upload, this fixes the root causes.
What you'll need
Step 1
Window → Workspaces → Audio. Verify Track Volume meter peaks at -6 to -3 dB during loud parts. Anything above 0 dB is clipping — fix at source.
Window → Workspaces → Audio. Premiere shows the Essential Sound panel and Audio Track Mixer.
Play the timeline. Watch the Track Volume meter (bottom right of monitor).
Target peaks: -6 to -3 dB during loudest parts (dialog peaks, music crescendos).
Clipping: anything above 0 dB. Indicated by red meter bars. Clipping is permanent distortion — cannot be fully fixed in post.
If source audio clips: best fix is re-record at lower input gain. If re-record impossible: use Essential Sound → Repair → De-Clip to reduce audible distortion.
Quiet source audio (peaks at -20 dB or below): boost via track-level gain or clip gain. Avoid boosting more than +6 dB without noise reduction; noise floor rises with signal.
Step 2
Effects → Audio → DeNoiser, DeHummer, DeReverb. Apply to dialog tracks. Manually edit out distracting breaths + clicks. Add room-tone bed under edits.
Cleanup pass before EQ/compression.
DeNoiser: Effects → Audio → DeNoise. Apply to noisy dialog. Default works well for hiss/hum/AC noise. Manual: Adaptive Noise Reduction effect for deeper control.
DeHummer: removes 50/60 Hz electrical hum from poorly grounded mics or AC interference.
DeReverb: reduces room reverb on dialog. Useful for podcasts/interviews in echoey rooms. Often introduces slight 'underwater' artifact — use moderately.
Manual breath edits: scrub through dialog, mute or reduce gain on distracting inhales/exhales between sentences. Time-consuming on long-form but produces clean dialog.
Room tone: record 30 sec of silence in the recording room. Loop on a low-volume bed track during dialog. Fills gaps where edits would otherwise sound 'dead.'
Click + pop removal: Effects → Audio → DeEsser (for sibilance) and manual edit for clicks (zoom in, find the spike, mute the single frame).
Step 3
Effects → Audio → Parametric Equalizer. High-pass at 80 Hz. Cut 200-300 Hz boxy. Slight boost at 3-5 kHz for presence. Verify with reference.
EQ shapes the frequency spectrum of voice to sit better in the mix.
Standard dialog EQ:
— High-pass filter at 80 Hz (removes low-end rumble that adds nothing to voice intelligibility)
— Slight cut at 200-300 Hz (-2 to -4 dB) — reduces 'boxy' resonance from many mics
— Slight boost at 3-5 kHz (+1 to +3 dB) — adds 'presence' and clarity to consonants
— Very slight boost or cut at 10-12 kHz depending on whether voice sounds dull (boost) or hissy (cut)
Apply EQ via Premiere's Parametric Equalizer or via Adobe Audition's effects (round-trip).
A/B compare: toggle the EQ effect on/off to verify it's improving clarity. EQ should be subtle — if it dramatically changes the voice character, you're overdoing it.
Step 4
Essential Sound → Loudness → Auto-Match. Or apply Multiband Compressor with -18 dB threshold + 3:1 ratio. Goal: dialog volume stays consistent.
Compression evens out quiet/loud parts of dialog so listeners don't constantly adjust volume.
Easy path: Essential Sound panel → select dialog clips → Loudness → Auto-Match. Premiere applies dialog leveling to match -14 LUFS (industry-standard YouTube/social) or -16 LUFS (podcast).
Manual path: Effects → Audio → Multiband Compressor. Settings:
— Threshold: -18 to -16 dB
— Ratio: 3:1 (compresses peaks 3:1 above threshold)
— Attack: fast (10 ms)
— Release: medium (100 ms)
Verify with meter: dialog peaks now sit around -3 to -1 dB consistently. Quiet parts boosted to -6 to -3 dB.
Don't over-compress: 6+ dB of gain reduction sounds 'squashed.' Aim for 2-4 dB of reduction on peaks.
Step 5
Music bed at -18 to -24 dB under voice. Use Audio Track Mixer → enable ducking, or manually keyframe music volume down during dialog.
Music should sit UNDER voice, not over.
Music bed level: -18 to -24 dB under dialog peaks. Music alone (no voice) can come up to -10 to -6 dB.
Auto-ducking: Premiere has built-in ducking. Essential Sound → Music → Auto Duck. Pick the dialog track as the reference. Premiere automatically lowers music volume when dialog plays.
Manual ducking: keyframe music volume on the timeline. Lower by 8-12 dB during dialog, restore between dialog. More control than auto.
Sound effects (SFX): place at the moment they happen on screen. Volume relative to scene — quiet ambient SFX at -30 to -25 dB, key SFX (impacts, transitions) at -10 to -3 dB.
Crossfade between music sections: drag default audio transition (Cmd+Shift+D / Ctrl+Shift+D) at edit points. Prevents abrupt music cuts.
Step 6
Window → Loudness Radar. Target -14 LUFS for YouTube/social, -16 LUFS for podcast, -23 LUFS for broadcast. Adjust master fader.
Loudness normalization is required for professional distribution.
Window → Loudness Radar (or use the Lumetri Audio panel meter).
Play the full timeline. Loudness Radar measures Integrated LUFS over the entire program.
Targets:
— YouTube / social video: -14 LUFS
— Podcast (Apple/Spotify spec): -16 LUFS
— Broadcast TV: -23 LUFS
— Cinema: -27 LUFS
If integrated LUFS is above target: lower master fader by the delta (e.g., +2 LUFS = lower master by 2 dB).
If below target: raise master fader OR apply gain to dialog tracks.
Don't aim too quiet (below -23 for social): platforms boost your audio relative to others, often introducing distortion.
Don't aim too loud (above -10 LUFS): platforms compress you down, also introducing artifacts.
Step 7
Listen on phone speaker, headphones, laptop speakers. Different devices reveal different issues. Fix what 80% of viewers will hear.
Final audio review before export.
Listen on at least 3 devices:
— Headphones (catches the most issues, but most viewers aren't on them)
— Laptop speakers (catches mid-range issues, common viewing setup)
— Phone speaker (catches whether dialog is intelligible on small speakers)
Common issues caught at this stage: dialog still uneven (re-compress), music too loud on phone (lower bed), sibilance ('ssss' sounds harsh on cheap speakers — apply DeEsser).
Fix issues then re-render the affected sections.
After export: verify the final file on phone + laptop. Don't trust the Premiere preview only.
If audio sounds different in the export than in Premiere: likely a sample rate mismatch (set export to 48 kHz) or an effect that didn't render (re-render the affected sequence).
Common mistakes
Clipped source audio
What goes wrong: Distortion baked into the source. No EQ/compression in post fully fixes it. Dialog sounds harsh and unprofessional.
How to avoid: Set input gain on the source side. Peak at -12 to -6 dB during loudest speech. Provides 6-9 dB headroom for unexpected peaks.
Skipping noise reduction
What goes wrong: Background hum, AC noise, room rumble all stay in the final. Audience trust drops; engagement falls 20-30%.
How to avoid: DeNoise, DeHummer, DeReverb pass on every dialog track. Catches 80% of common issues in 5-10 min.
No EQ on dialog
What goes wrong: Voice sits in the wrong part of the spectrum. Sounds muddy or thin. Music + dialog clash for frequency space.
How to avoid: Standard dialog EQ template: high-pass 80 Hz, slight cut 200-300 Hz, slight boost 3-5 kHz. Save as preset. Apply to every dialog track.
Music too loud
What goes wrong: Music drowns dialog. Listeners turn down the whole video. Engagement drops because dialog is the content.
How to avoid: Music bed -18 to -24 dB below dialog peaks. Use auto-ducking or manual keyframing. Verify in Loudness Radar.
No loudness normalization
What goes wrong: Video uploads, platform re-normalizes, audio quality degrades. Video sounds thin or harsh after platform processing.
How to avoid: Master to -14 LUFS for YouTube/social, -16 LUFS for podcast. Loudness Radar shows the integrated LUFS; adjust master fader to hit target.
Mixing on laptop speakers
What goes wrong: Bass-heavy mix that sounds great on laptop but boomy on phone, thin on TV, harsh on headphones. Audio doesn't translate across devices.
How to avoid: Mix primarily on neutral headphones (Sony MDR-7506, AKG K240, Audio-Technica M50x). Verify on 3 other devices before final export.
Recap
Done — what's next
How to set up an Adobe Premiere Pro project correctly
Read the next tutorial
Hand it off
Audio mixing takes 30-60 min per video to do well. For weekly content with professional polish, hiring a video editor with audio chops is the right move. EverestX matches at $14-16/hr — typically $400-1,000/mo for ongoing audio polish on top of editing.
See video editor rates
YouTube and social video: -14 LUFS integrated. Podcast (Apple, Spotify): -16 LUFS integrated. Broadcast/TV: -23 LUFS. Cinema: -27 LUFS. Match the target — platforms re-normalize if you're off, sometimes introducing distortion.
Three usual causes: (1) sample rate mismatch — set export to 48 kHz, (2) an effect didn't render before export — re-render the affected sequence, (3) the export uses a different output configuration — check Sequence Settings → Audio match the export. Verify by importing the export back into Premiere and comparing.
For most marketing/podcast/YouTube workflows: Premiere is sufficient. The Essential Sound panel handles 80% of needs. Send to Audition for: deeper noise reduction (Adaptive Noise Reduction), multi-track mixing with sub-mixes, advanced surround sound, or commercial broadcast work. Round-trip via Edit Clip in Adobe Audition.
Compress each speaker's track independently to match levels. Or use Essential Sound → Dialog → Auto-Match Clip Volume, which matches all selected clips to a common loudness. Manual: clip-level gain (Cmd+L / Ctrl+L on a clip) to bring quiet clips up before compression.
Sibilance is amplified by some codecs (lossy compression accentuates high frequencies). Apply Premiere's DeEsser (Effects → Audio) before compression. Target the 6-8 kHz range with a 3:1 reduction. Small reduction (1-2 dB) usually does the trick without making voice sound dull.
Adobe Premiere Pro
Premiere Pro is forgiving for the first hour and brutal in the second. Skip the project setup and you get crashes, lost work, slow scrubbing, and exports that take all night. Here's the setup that actually keeps Premiere fast and stable.
Adobe Premiere Pro
Multi-cam is what separates Premiere from CapCut and Descript. Two cameras on a podcast guest, three on an interview, six on a concert — sync them in 30 seconds, cut live like a switcher, deliver in half the time. Here's the full workflow.
Adobe Premiere Pro
Lumetri is Premiere's color tool — and the difference between amateur and professional video. This walks the full workflow: scopes, primary correction, creative looks, skin-tone protection, and the per-clip vs adjustment-layer decision that most DIY editors get wrong.
Adobe Premiere Pro
Most teams hit the DIY Premiere ceiling at 10-30 projects. Quality plateaus, the learning curve never ends, and production cadence slips. Here's the honest framework for when to hire a professional video editor — and what that role actually does.
Descript
Most podcasters bounce between 4 tools: Riverside for recording, Descript for editing, Audacity for cleanup, Canva for show art. Descript can do 80% of it in one app. Here's the full episode workflow.