Video summary

How to Use Audio Compression for Post-Production: Film, TV, Podcasts & YouTube Explained *LIVE*

Main summary

Key takeaways

Technology

Summary of the Video (Audio Compression for Post-Production)

The live stream explains how to use audio compression in post-production for film/TV dialogue, podcasts, and YouTube, with hands-on demonstrations in Pro Tools. The speaker emphasizes that compression is harder to learn by ear than volume changes, because it alters dynamic range by reducing peaks vs. dips, and because its behavior is time-based via attack and release.

Demonstration Setup (Pro Tools + Voice/ADR/Foley)

The speaker records voice/ADR and routes audio using an “AX” input style channel strip concept (commonly used in ADR/Foley mixing). They add compressors/limiters in Pro Tools and monitor results using:

  • Gain reduction meters (how much compression is happening)
  • Output/short-term loudness readings (noting YouTube preferences)

Why Compression Is Used

  • Core purpose: reduce dynamic range so dialogue sounds more polished/professional and remains more consistent
  • The speaker discusses two main uses:
    • Taming dynamics (practical leveling)
    • Creative shaping of how a voice sits in the mix (including mix “glue”)
  • They also argue you shouldn’t blindly follow common forum advice—use your ears.

Compressor Controls Covered (and Their Effects)

1) Makeup Gain (Output Compensation)

Makeup gain restores level after compression reduces peaks. For YouTube, the speaker notes a common target is to keep average loudness around roughly -16 to -4, based on platform expectations.

2) Threshold + Knee (When Compression Starts)

  • Threshold determines when the compressor kicks in
  • Knee smooths or sharpens the transition:
    • Low/0 knee = sharp cutoff (compression begins right after threshold)
    • Higher knee = compression starts below threshold more gradually
  • They recommend a moderate knee for dialogue to avoid over-processing quiet material.

3) Ratio (How Much Gain Reduction Happens)

  • Ratio controls how aggressively the signal is reduced once it crosses threshold
  • Examples:
    • 1:1 = effectively no compression
    • Higher ratios clamp harder (they demo settings causing reductions like ~15 dB)

4) Attack (How Quickly It Reacts)

Attack is critical for dialogue intelligibility.

  • Very fast attack can clamp down on consonant detail
  • The speaker references research suggesting faster attack can reduce audience ability to understand speech

Practical starting points:

  • For dialogue: ~5–10 ms attack
  • For bus compression: attack can be slower for a more gentle leveling feel

5) Release (How Quickly It Stops Compressing)

Release controls how long gain reduction continues after the signal drops.

  • For dialogue, they prefer roughly 100–200 ms
  • Release timing is also useful for sidechain/ducking, such as ducking music under voice

6) Depth (Max Compression Amount, Where Supported)

Some compressors include depth/limit of gain reduction to prevent over-clamping, letting louder moments still cut through. They recommend depth-like control for film/TV dialogue where loud moments and music/SFX coexist.

7) Dry/Wet (Blend Compressed vs. Uncompressed)

Dry/wet lets you blend heavy compression with some natural dynamics.

  • It interacts with makeup gain—if makeup gain is high, dry/wet differences may be less noticeable.

Rules-of-Thumb Challenged

The speaker critiques a common forum suggestion like “3–6 dB of compression.” Their position:

  • It may be appropriate depending on material and mix goals
  • Heavy compression everywhere can make mixing feel like “everything is graded”
  • Compressing one element more heavily (while leaving others dynamic) can help elements move through the mix faster

Noise Floor + Dialogue Compression (Important Caveat)

Compression can raise the perceived noise floor because it reduces louder portions and makes quiet background content more audible.

Example mentioned:

  • As compression and makeup gain increase, fan noise becomes more noticeable

For dialogue, they generally prefer less processing than many typical mixes to avoid “nuking” ambience—though they acknowledge noise reduction can help.

Q&A Topics / Workflow Guidance

Track vs. Bus Compression (and Multistage Approaches)

  • They generally avoid too much track compression unless turnaround time forces it
  • Preferred workflow is often:
    • bus compression to create smoother, softer glue when multiple sources (e.g., boom + lavs) contribute simultaneously
  • If stacking stages (channel + bus):
    • the bus should typically be the gentler/softer stage

Handling Compressed Production Dialogue (Already Squashed Inputs)

  • It’s very difficult to “uncompress” audio that’s heavily compressed/limited
  • If audio is clipped, tools like de-clipping may help somewhat
  • But if dynamics are already collapsed, expanders/upward compression can’t truly restore original dynamics
  • They cite an example of a wireless mic whose output sounded “super compressed,” suggesting you can’t fully undo it in post

Matching Boom and Lav Mics

General observation:

  • lavs are often more compressed (mic/transmission processing)
  • boom mics are usually less compressed

Recommended approach:

  • Use bus compression to “glue” them together rather than treating each mic identically with aggressive settings

Whispering / Broadcast Spec Delivery

  • For whisper dialogue, you may need to raise whisper volume to meet loudness/spec
  • They discuss differences:
    • onscreen whispering: lip movement supports intelligibility
    • cutaways/B-roll: lipread cues disappear, requiring more careful level management

Film vs. TV Documentary Compression Differences

They argue modern viewing reduces extreme differences, but:

  • films often preserve more dynamic range (tradition/aesthetics)
  • TV/home viewing tends to demand more consistent loudness

Multiband Compression for Dialogue/SFX (Advanced)

They recommend handling resonance problems via dynamic EQ/compression, such as:

  • using dynamic EQ behavior (similar to a Pro-Q-like workflow) to compress only problematic bands

Example:

  • resonance around ~400–500 Hz (noted from a specific mic/transducer example)

They position this as an alternative to dedicated multiband compressors when you want targeted control.

Plugin Announcements + Development Demos (Major Product Content)

“DXB Channel Strip” Plugin (Commercial Release, Mac First)

A dialogue-focused channel strip concept including:

  • EQ (highpass, lowpass, and high-frequency gain)
  • transparent compressor (built-in)
  • DSER (de-ess/sibilance control)
  • safety limiter at the end (non-user toggleable)

Claims/features:

  • Designed to be hard to overdrive/clip easily
  • About ~8–10% CPU use (noted as a downside)

Availability:

  • $30
  • Mac version first (PC version work mentioned)

GUI note:

  • Described as “boutique/grungy” visually, but focused on sound.

“AutoComp” Plugin (Automatic Compressor Prototype, Near Release)

A “dream” compressor concept:

  • automatic threshold + automatic makeup derived from detected input

Controls are intentionally minimal:

  • Compression amount (percentage)
  • Output gain
  • Option to turn a limiter on/off

Conceptual behavior:

  • Threshold derived from detected RMS
  • Makeup calculated from detected gain reduction/RMS behavior
  • Includes a dynamic low-volume boost so quiet speech stays “in the pocket”

Practical use:

  • Dialogue consistency for broadcast/YouTube/live without manual threshold/ratio tuning

CPU/performance:

  • Said to be lightweight (contrasted with DXB being CPU-hungry)

Testing plans:

  • Wants channel members to help test before wider release

Release timing:

  • Speaker suggests within ~2 weeks for near-release stage

Mentioned Sources / Main Speakers

  • Main speaker: “Brandon” (host of the stream)
  • Primary software/tooling: Pro Tools (Avid)
  • Referenced plugin/software families/tools:
    • FabFilter
    • Pro-Q3 / Pro-Q
    • iZotope dClipper
  • Frequent chat contributors referenced: “Simon” and others (Oliver/Olivier, Israel Hernandez, Hans, House of Jen, TJ, Master Nelson, etc.)

Original video