Video summary

Dialogue EQ Secrets for Film & YouTube

Main summary

Key takeaways

Educational

Main Ideas, Concepts, and Lessons (EQ Dialogue)

1) Why EQ Dialogue Is Necessary (Creator + Mixer Perspective)

  • Raw dialogue recordings usually need EQ to sound full, rich, clear, and professional.
  • EQing dialogue is harder than it seems because you must be able to hear what’s wrong—specifically, identify frequency balance deficiencies.
  • EQ should be about moving from point A to point B, not applying presets or doing “random EQ.”
  • Editing affects EQ workload:
    • A good dialogue edit (tight cuts, consistent continuity) reduces problems caused by microphone placement, mic choice, and room tone changes.
    • A poor edit forces you into more drastic EQ/noise reduction later.
  • EQ can be done with any tool (stock or paid): the method is what matters.

2) Approach: Work Frequency Bands From Low → Mid → High

The speaker frames corrections by frequency region:

  • Lows: rumble/mud
  • Mids: boxiness, phasing, harshness around intelligibility
  • Highs: ice-pick harshness + intelligibility without sounding brittle

Core principle: use your ears—because every mic/room/actor differs.

Detailed Methodology / Step-by-Step

A) Low-Frequency Problem Solving (Boomy/Muddy Dialogue)

Common Symptoms

  • “Too boomy” / “too muddy”
  • Often caused by overemphasized low frequencies, not by missing highs.

Core Fixes

  • High-pass filter (HPF / low-cut) first

    • Purpose: remove rumble/hum/sub-bass that doesn’t support dialogue clarity.
    • Avoid “eating into” the speaker’s fundamentals:
      • Example: if a fundamental is around ~70 Hz, raising the HPF to ~100 Hz with a 12 dB/octave slope can reduce mud while preserving tone.
    • Prefer gentle slopes (6 or 12 dB per octave) instead of brick-wall cuts.

  • Cut setting guidance

    • For many creators/podcasters, you may need to cut higher than you expect since viewers don’t need sub-bass in spoken dialogue.
    • Example guidance: could be as high as ~200 Hz depending on taste/content.

Why It Matters (Beyond Tone)

  • Viewer devices: phones/tablets/TV make sub-bass unnecessary and potentially problematic.
  • Headroom/bit-depth analogy: carving lows creates usable space so mids/highs get more of the “detail budget” in the digital file.
  • Plosives: HPF can reduce some “P” pops close to the mic (proximity effect + no pop filter).

Optional Pro-Level Enhancement: HPF + Low Shelf

  • Combine:
    • HPF to remove mud
    • Low shelf to restore upper bass body (often ~100–200 Hz)
  • Use caution: too much boost can cause resonances.
  • Goal: sound less thin even after reducing lows.

If You Want Perceived Low End Back (Thin Clips)

Tools mentioned:

  • Low-target saturation (iZotope workflow noted; in Pro Tools saturation may be more full-band unless specialized tools are used)
  • Subharmonic generation (warning: can sound artificial like LFE if overdone)
  • Example: Waves R-Bass for perceived low end without full rumble.

Pro Tools Practical Suggestion

  • Starting chain example:
    • 6 dB/octave HPF
    • Optionally a low shelf near 0 initially, then adjust
    • Use automation for scene/segment changes

B) Midrange Cleanup (Trickiest Area—Especially for Lavaliers)

Typical Issues

  • Mids are described as the most difficult band.
  • Lav mic placement is often the primary offender:
    • Capsules often sit near the zipper/chest (jaw/chin “shadow” effects).
    • This can cause phasiness or strange midrange behavior.

Where to Look

Common problem areas (not “always”):

  • Around ~200–500 Hz
  • Or sometimes ~180–250 Hz or ~400–600 Hz

The guidance is to find what’s offensive for that mic/actor/costume.

Correction Technique

  • Use small notch dips only where needed.
  • Workflow trick:
    • Loop a problematic phrase, raise gain to find the resonant frequency, then dip by a small amount (often 1–2 dB).
  • Avoid over-EQing:
    • Don’t create an EQ curve with lots of boosts/cuts.
    • Over-correction can create artifacts (the speaker references “hair” / comb-filter-like issues and asks viewers to identify a specific artifact).

When Combining Mics (Shotgun + Lav)

  • If shotgun + lav are both used:
    • The lav may require less mid EQ because it supports the shotgun/broadcast sound.
  • If it’s lav-only:
    • Expect more cleanup.

Don’t Change “Identity”

  • Don’t remove a speaker’s character just to match other voices—clarity should be balanced with natural timbre.

Optional: Boosting Mids (Less Common)

  • Sometimes appropriate if a mic lacks mids and you need dialogue clarity/power.
  • Risks:
    • Avoid making lavs honky
    • Avoid amplifying unwanted resonances
  • Mentioned: some large diaphragm condensers may benefit from slight mid boost because cheaper LDCs can sound “smiley” (too much high/low relative to mids).

Advanced/Optional Tools

  • Dynamic EQ / multiband dynamics

    • Use sparingly for resonances that poke out on specific words.
    • Approach referenced in FabFilter: right-click to make dynamic and choose comp/expansion behavior.

  • EQ matching

    • Harder in mids due to multiple sub-regions (woodiness vs upper mids).
    • Mentioned example: “upper mid bite” around ~1–2 kHz; sometimes a small cut (e.g., ~3 dB) removes bite (example: nasal female voices).

  • Mid/side EQ

    • Dialogue is mostly treated as mono, so mid/side EQ is generally less applicable, but discussion is welcomed.

C) High-Frequency Handling (Clarity Without Harshness)

Common Creator Pitfall

  • Over-boosting highs until they become ice-pick harsh.
  • On phones, excessive treble can sound cheap/tearing.

Core Approach: Gentle Low-Pass + Targeted Clarity Boost

  • Use low-pass (opposite of HPF) carefully to control excessive top-end, including hiss/wireless artifacts.
  • Don’t set cutoff too low or it can turn muddy.
  • Suggested starting point:
    • Low-pass around ~12–14 kHz with gentle slopes

“Sweet Spot” Concept

  • With gentle slopes, lowering the cutoff doesn’t instantly “muffle everything” because attenuation is gradual.

Best Trick Described: High Shelf Instead of Wide High Boost

  • Instead of boosting the entire high spectrum:
    • Use a high shelf for subtle intelligibility lift
  • Intended effect:
    • Improves clarity mainly in the ~2 kHz–5 kHz intelligence region
    • Reduces harsh hiss without boosting everything above.

Avoid Extreme High Shelves

  • Too much boost becomes harsh and fatiguing.

De-Essing (Sibilance Control)

  • Treated as dynamic EQ for sibilance.
  • Warning: too much de-essing can make S’s sound like lisps or produce crunchy artifacts.
  • Alternatives/workflows mentioned:
    • FabFilter de-esser praised; Dyn3 called less effective
    • iZotope syllable-by-syllable editing / spectral repair for gentle control

Recommendation:

  • De-ess very gently—fix only what sticks out.
  • If whistle-s sounds appear, spectral repair can work well (tame rather than remove everything).

Monitoring Guidance

  • Speakers are preferred over headphones for EQ decisions because headphones miss the “air space” between driver and ear.
  • If using headphones:
    • Ensure they’re flat (calibrated/flat response), or don’t rely on them for EQ.

Consistency Rules (So Dialogue Sounds the Same Across Shots)

  • Tonal consistency should come from:
    • Good dialogue editing (smooth scene tone transitions)
    • Avoiding too many obvious EQ changes that “call attention” to shot transitions
  • General rule:
    • Dialogue should remain consistent shot-to-shot to preserve performance/story connection
  • Exception:
    • Intentionally different scenes/lines (e.g., villain close-up delivering an epic low line) may justify more noticeable EQ/level/tone changes.

Philosophy: Avoid “Over-EQ”

  • Over-EQ is a major problem.
  • Mindset:
    • Start flat → listen → make fewer, smaller decisions
  • Each processing pass can degrade quality:
    • compression, EQ, noise reduction can add artifacts
  • You can’t add missing frequency content that wasn’t recorded:
    • EQ can emphasize what exists, but can’t truly restore what’s absent
  • Summary idea: Do less processing, do more decision-making.

Film/Music Comparison (Why Dialogue EQ Differs From Music)

  • The speaker criticizes music-style “bright/icy” vocal chains used in studios (e.g., aggressive boosts over 10 kHz).
  • Dialogue/video mixing should prioritize:
    • natural, pleasant sound for long viewing sessions
    • avoiding ear fatigue
  • In film/TV/YouTube:
    • avoid super shrill, highly exciter/saturated treble approaches common in some music contexts.

Plugin/Utility Announcement: Auto Level (End-of-Stream “Dessert”)

What It Does

  • “Auto Level” automatically sets audio loudness to a target output level.
  • It does not compress:
    • it measures per-sample/level behavior and adjusts gain to match loudness.

Intended Use

  • YouTube, podcasts, dialogue leveling at scale (e.g., video games with many lines)
  • Live streams needing consistent spec-compliant levels

User Control

  • Essentially one control: level target (input/output gain goal)
  • Example targets:
    • YouTube/podcast-like: around -15 output
    • TV spec example: around -24 output

Technical Notes

  • Includes a limiter stage (hard limiting to prevent overs)
  • Latency judged low enough for lip sync tests
  • CPU usage reported around ~5–12% during testing (on an M2 Ultra)

Release and Pricing Expectations

  • Planned release: within 1–2 weeks
  • Suggested price: around $20
  • Channel members get early installer/testing access
  • Likely formats discussed:
    • AU and VST (AAX/Pro Tools version may be optional)

Speakers / Sources Featured

People

  • Tom (primary speaker; host of the livestream)
  • Ty Sonic Production (mentioned as a new channel member; not a direct speaker in the transcript)
  • Other names appear as chat commenters/questions, including:
    • Tiziano, Umut Bars, Lake One Sound, Kevin Broman CAS, Sasha, Chase, Simon, Greg, Brandon, Javier, Queen’s Gambit mixer, Obi, and others

Tools / Plugins / Systems Referenced

  • iZotope Insight / Insight 2 (spectrogram/loudness analysis)
  • iZotope Spectrum
  • Pro Tools
  • FabFilter Pro-Q 3
  • FabFilter DNS 2
  • FabFilter de-esser
  • iZotope syllable editing / spectral repair
  • Waves R-Bass
  • Waves Saturn (specialized saturation use)
  • SoundID Reference (calibration option)
  • Eventide split EQ (asked about, not used)
  • Pro Sub / subharmonic generator
  • Pro Multiband Dynamics
  • Dyn3 (Pro Tools de-esser referenced as not ideal)
  • DaVinci Resolve templates/practice projects
  • Pro-Q 3 / EQ3 control behaviors (as referenced during solo/band focus)

Recording / Brand References Mentioned

  • Morgan Freeman, Taylor Swift, Stevie Nicks, Michael Jackson
  • DJI mic (example of too-close recording)
  • Rode NT1, Neumann, AKG C414
  • Mentions of 32-bit recordings, field recording, and “guns” (future content)

Original video