Premiere Pro Beginner Workflow: Audio Basics for Clear Dialogue and Balanced Mix

Audio concepts you must understand (in practical terms)

Levels: what the meters are telling you

Premiere’s audio meters show signal level in dBFS (decibels relative to full scale). In digital audio, 0 dBFS is the ceiling. Everything is measured as negative numbers below that ceiling (for example, -12 dBFS).

• Dialogue sweet spot (typical): peaks around -6 to -3 dBFS, with average speech often living around -18 to -12 dBFS before loudness normalization.
• Music bed: usually much lower than dialogue while someone is speaking, often -25 to -15 dBFS depending on the track and the voice.

Clipping: why “red” is a problem

Clipping happens when audio tries to go above 0 dBFS and gets flattened. It sounds harsh and crunchy, and it’s difficult to fix. In Premiere, watch for red meter peaks and the clip/track meter indicators.

• If a dialogue clip is clipping, reduce level before it hits the ceiling (clip gain is usually the first fix).
• If it’s already recorded clipped, you may reduce how noticeable it is, but you can’t truly restore the missing waveform.

Noise floor: the “room tone” underneath everything

Noise floor is the constant background sound in a recording (HVAC, computer fan, street noise). If you boost dialogue too much, you also boost the noise floor. The goal is not “silence,” but consistent, unobtrusive background that doesn’t distract.

Mono vs stereo: choose the right channel behavior

• Mono dialogue is common for lavs/shotguns. It should usually play equally in left and right (centered).
• Stereo music should remain stereo to keep width and depth.
• Dual-mono recordings (two mics recorded to L/R) can be mistaken for “stereo.” If one side is the good mic and the other is noisy, you’ll hear imbalance unless you fix channel mapping or convert to mono.

Track types in Premiere: what they’re for

• Standard audio tracks: most common for dialogue, music, and SFX.
• Submix tracks: route multiple tracks into one “bus” so you can process/control them together (useful for a Dialogue submix and a Music submix).
• Master track: final output stage; keep it clean and avoid heavy processing unless you know why you need it.

Dialogue-first workflow (fast and reliable)

This workflow prioritizes intelligible speech first, then supports it with music and effects. You’ll use Essential Sound to apply sensible processing quickly and consistently.

Step 1: Set up a simple audio layout

1. Put dialogue on its own track(s) (for example, A1 = Dialogue).

   Continue in our app.

   • Listen to the audio with the screen off.
   • Earn a certificate upon completion.
   • Over 5000 courses for you to explore!

   Or continue reading below...

   

   Download the app

2. Put music on its own track (A2 = Music).

3. Put SFX/ambience on their own track (A3 = SFX/Ambience).

4. Open Window > Essential Sound.

Step 2: Tag audio with Essential Sound (Dialogue/Music/SFX)

1. Select your dialogue clips in the timeline.

2. In Essential Sound, click Dialogue.

3. Select music clips, click Music.

4. Select effects/ambience, click SFX (or Ambience if appropriate).

Tagging does two things: it gives you relevant controls (like Reduce Noise for dialogue), and it helps you stay consistent when you apply similar settings across many clips.

Loudness targets for common web delivery

There are two ways people talk about “how loud” audio is:

• Peak level (dBFS): momentary maximums (good for avoiding clipping).
• Loudness (LUFS): perceived average loudness over time (good for consistent playback across devices).

Also keep true peak under control (commonly ≤ -1 dBTP). If you don’t have true peak metering available, a practical beginner rule is: keep peaks below -1 dBFS and avoid any red clipping indicators.

Clean up dialogue with Essential Sound (practical settings)

Step 1: Fix level problems first (Clip Gain before Volume)

If a dialogue clip is too quiet or too loud compared to the rest, adjust Clip Gain so your fader/volume automation stays reasonable.

1. Select the dialogue clip.

2. Right-click > Audio Gain…

3. Use Adjust Gain by to raise/lower in small steps (try ±2 to ±6 dB).

4. Play the loudest words and watch meters; aim for healthy peaks without hitting red.

Why this matters: Clip Gain changes the level before effects and processing, which helps compressors/noise reduction behave more predictably.

Step 2: Use Essential Sound > Dialogue for clarity

Select a dialogue clip (or a group), then in Essential Sound (Dialogue):

• Loudness: enable Auto-Match to quickly even out dialogue clips. Use it as a starting point, then fine-tune by ear.
• Repair:
  • Reduce Noise: start low (e.g., 2–4). Increase until noise is less distracting, but stop if the voice starts sounding watery/phasey.
  • Reduce Reverb: use gently (e.g., 1–3). Too much can create artifacts and make speech dull.
  • DeHum: enable if you hear a steady electrical hum (50/60 Hz). If it doesn’t help, turn it off.
• Clarity:
  • EQ: enable and choose a preset like “Vocal Presence” as a starting point.
  • Dynamics: use light compression to keep speech consistent. If it starts pumping or bringing up noise between words, back off.

Basic EQ moves (simple and safe)

If you open the EQ (from Essential Sound or the effect controls), these are common beginner-friendly moves:

• High-pass filter (low cut): roll off rumble below ~70–100 Hz for most voices (higher for thin voices, lower for deep voices). This often cleans up HVAC/handling noise.
• Presence boost: a gentle lift around ~3–5 kHz can improve intelligibility. Keep it subtle to avoid harshness.
• Reduce muddiness: if the voice sounds boxy, a small cut around ~200–400 Hz can help.

Make small changes (1–3 dB) and A/B frequently (toggle the effect on/off) to ensure you’re improving, not just changing.

Noise reduction: what “too much” sounds like

Overdoing noise reduction often creates:

• Swirly/underwater artifacts
• Choppy ambience between words
• Lispy consonants

If you hear these, reduce the amount and consider leaving a little consistent room tone rather than forcing silence.

Balance music under dialogue (without fighting the voice)

Step 1: Set a rough music bed level

1. Play a section with dialogue and music together.

2. Lower the music track/clip volume until you can understand every word without effort.

3. As a starting point, music during speech is often 10–20 dB lower than the dialogue, but always trust intelligibility over numbers.

Step 2: Use Essential Sound > Music “Ducking” (quick method)

1. Select the music clip(s) and tag as Music in Essential Sound.

2. In Essential Sound (Music), enable Ducking.

3. Set Duck Against to your dialogue track (or “Dialogue” if available).

4. Click Generate Keyframes.

5. Adjust:

   • Reduce By: how far music dips (try 8–16 dB).
   • Sensitivity: how easily it detects speech (raise if it misses words; lower if it ducks too often).
   • Fades: smoothness of transitions (longer fades sound more natural).

After generating, you can manually tweak the resulting keyframes if a word still gets masked or if the music dips too aggressively.

Fix problem moments with volume keyframes (manual control)

Even with Auto-Match and Ducking, you’ll often need small manual fixes: a laugh that spikes, a word that drops, or a music hit that steps on a sentence.

Step-by-step: keyframe clip volume for a single moment

1. In the timeline, expand the audio track height so you can see the volume line clearly.

2. Use the Pen tool (or Ctrl/Cmd-click on the volume line) to add four keyframes: two before and two after the problem area.

3. Drag the middle segment up/down to raise or lower just that moment.

4. Play through the edit point and listen for smoothness; adjust spacing between keyframes to make the change more gradual.

When to use Clip Gain vs keyframes

• Clip Gain: use when the entire clip is consistently too loud/quiet or when you need to feed effects a better level.
• Keyframes: use for short sections inside a clip (one phrase, one breath, one music hit).

Mini-mix checklist (quick pass before export)

• No clipping: scan loud sections; ensure meters never hit red.
• Dialogue intelligibility: you can understand every sentence on small speakers at moderate volume.
• Consistent dialogue level: no sudden jumps between lines or shots; Auto-Match plus small keyframe tweaks.
• Noise is controlled: background is not distracting; noise reduction is not creating obvious artifacts.
• Music supports, not competes: music dips under speech; transitions are smooth (no pumping).
• Room tone continuity: no unnatural dead-silent gaps between phrases unless intentionally muted.
• Headphone check: listen for hums, clicks, mouth noises, and harsh sibilance (S sounds).
• Phone/laptop check: confirm speech remains clear on limited speakers.

Assignment: clean dialogue, balance music, export a mix test

Goal

Create a 20–40 second “mix test” where dialogue is clean and consistent, music sits underneath without masking words, and the output is safely below clipping.

What you need

• One dialogue clip with some background noise (or room tone)
• One music clip (stereo is fine)

Steps

1. Tag clips in Essential Sound: dialogue as Dialogue, music as Music.

2. Dialogue cleanup:

   • Adjust Clip Gain so peaks are healthy and not near clipping.
   • In Dialogue settings, enable Auto-Match.
   • Apply Reduce Noise gently; add EQ with a light high-pass filter if needed.

3. Music balance:

   • Lower music volume until dialogue is effortless to understand.
   • Enable Ducking on the music and Generate Keyframes.
   • Manually adjust keyframes where the music still competes with key words.

4. Problem-moment fix: find one spot where a word is too quiet or too loud and correct it with volume keyframes (four-keyframe method).

5. Mix test export: export a short file (20–40 seconds) and listen on headphones and phone speakers. If speech is hard to understand on the phone, reduce music further or add a small presence boost to dialogue.

Self-check questions

• Do any words disappear when music is playing?
• Do you hear pumping or underwater artifacts from noise reduction?
• Does the loudest moment stay below clipping?
• Is the dialogue level consistent from the first line to the last?