Video summary

The Ultimate Dialogue and Vocal Mixing Channel Strip: Compression, EQ, and De-Essing

Main summary

Key takeaways

Product Review

Product Overview: DXB2 Dialogue/Vocal Channel Strip

The video reviews DXB version 2 (“DXB2”), a custom-built dialogue and vocal mixing channel strip plugin made by an individual developer. The main idea is that it replaces a typical 4–6 plugin dialogue bus chain with one efficient plugin.

What It Replaces (Typical Old Workflow)

The creator says their dialogue bus often uses 4–6 plugins, commonly including:

  • EQ
  • Bus compression
  • De-essing
  • Limiter
  • Loudness monitoring (e.g., LUFS/Insight)

DXB2 is positioned as replacing ~5 plugins with a single plug-in.

Main Features (DXB2)

1) Channel-Strip-Style Processing Built Around EQ

DXB2’s UI is designed for an EQ-first tonal workflow, including:

  • High-pass
  • Low shelf
  • Three mid bands (with adjustable gain, filter, and Q)
  • “Air” band for smoother high-frequency boost/cut
  • Low-pass
  • Reset behavior via double-click
  • Resizable UI; closing the UI improves efficiency

2) Frequency-Aware De-Esser Integrated into the EQ (“Air” Band)

The orange line in the air band is described as a frequency-aware de-esser:

  • Adjust via right-click + drag on the orange line
  • Double-click resets to “no de-essing”
  • Claims low system usage while running (example: ~3%)
  • Styled so EQ and de-essing are handled together

3) Compressor Section Using Integrated AutoComp

DXB2 includes an AutoComp compressor built into the plugin:

  • Described as transparent
  • Uses RMS compression, smart threshold, and makeup gain
  • Compression can be pushed up to 100%
  • Features a Lift control (dynamic gain for quieter moments)

Lift control’s purpose: Helps prevent the common compressor issue where quieter parts (whispers, low levels, live streaming drops) fall below the threshold and receive little or no compression. Lift can be toggled on/off via right-click.

4) Built-In Loudness Workflow (Cheap LUFS Readout)

DXB2 provides:

  • Input LUFS
  • Output gain trim

It targets a TV-style -24 LUFS workflow:

  • When output gain is 0, the creator says it targets around -24 LUFS
  • For a YouTube-style approach, output gain can be boosted (example: +8 dB) to land around -16 to -18 LUFS

5) Limiter with “Quasi-Peak” Behavior (Zero Latency)

DXB2 includes a limiter stage that is:

  • Zero latency (not a look-ahead true peak limiter)
  • Framed as “as close as possible” while remaining suitable for live/broadcast
  • Switchable off (right-click)

If the limiter is turned off, it effectively allows clipping behavior—so caution is implied.

6) Presets + “Pill” Style Toggles

  • Supports presets
  • Uses “pill” controls where right-clicking toggles options like:
    • Limiter
    • Lift

System Performance & Latency Comparisons

DXB2 Latency

  • Reported as ~960 samples (~20 ms) for the DXB2 + noise reduction test.

Old Multi-Plugin Chain Latency

  • Compared chain reportedly reached up to 8,443 samples (“a lot,” with ms not explicitly calculated).

Verdict from the video: latency is much improved with DXB2 and it’s considered usable for live/broadcast, with automation still possible.

Noise Reduction Integration: Where to Place It

The video focuses on how to place noise reduction relative to DXB2.

Common Advice vs. Creator’s Preferred Approach

  • Common advice: place noise reduction post-EQ and before compression
  • Creator’s workflow (especially with Voxengo):
    • Place Voxengo noise reduction BEFORE the channel strip (before DXB2)

Stated reason: next-gen NR plugins (including Voxengo) often work better with full-range audio and may not use AI reconstruction the same way others do.

Technical Reasoning (As Stated)

  • Voxengo uses a filter bank approach (not AI/neural-net)
  • The creator claims it still benefits from receiving noisier / full-range input
  • Heavy AI-based NR approaches are also suggested to work better with full-range audio

Practical Benefit Claimed

  • If the input is already clean, Voxengo won’t do much (little noise to reduce)
  • On noisier sources (traffic/HVAC/waterfalls), more noise = more reduction

Potential Issue and Fix

If NR dulls highs too much:

  • Use DXB2’s Air/EQ to restore clarity.

Pros Mentioned / Implied

  • Streamlines workflow: replaces ~4–5 plugins with one
  • Dialogue-focused: built for spoken voice use cases (YouTube, films/TV, podcasts, live streaming)
  • Sound quality: described as great on dialogue and vocals and transparent compression
  • Efficiency: “super efficient DSP stack,” with de-esser example ~3% CPU
  • Intuitive EQ-first UI for fast tonal adjustment
  • Integrated modules reduce plugin juggling:
    • EQ, de-essing, compression, limiting, loudness-oriented trimming
  • Usable for live/broadcast due to zero-latency design philosophy
  • Lift control addresses threshold-related “no compression” problems

Cons / Limitations Mentioned

  • No “fancy marketing/polish”: UI described as utilitarian, not as visually refined as major brands
  • Mentions that DXB1 had an ugly UI and inefficient DSP (history behind the current rebuild)
  • Limiter is zero latency but not true peak look-ahead, so it’s not positioned as the most aggressive peak limiter style
  • Turning the limiter off may allow clipping

Comparisons Made

  • FabFilter: praised for visual beauty/quality, though DXB can’t compete on “fanciness”
  • Avid/Pro Tools channel strip UI: considered better for deep tweaking, while DXB aims for an EQ-like intuitive workflow
  • Pro-C 2: mentioned as an example of the “threshold not being hit = no compression” issue DXB’s lift/smart threshold tries to solve
  • Noise reduction routing: traditional routing vs. creator’s Voxengo-before-strip method
  • Latency: DXB2 (~960 samples) vs older multi-plugin chain (up to 8,443 samples)

User Experience & Workflow Notes

  • Built to be fast to dial in
    • EQ bands are front-and-center
    • Double-click resets
  • De-esser demonstration includes live streaming context
  • DXB2 can be resized, and closing the UI improves CPU efficiency
  • Emphasizes reducing the need to reorder plugins when adding noise reduction

Verdict / Recommendation

DXB2 is recommended if you want a faster, more efficient dialogue/vocal workflow, especially for spoken-word content. It consolidates EQ + de-essing + compression + limiting + loudness trimming into one plugin with low latency suitable for live/broadcast.

The trade-off highlighted in the video is that the interface is more utilitarian than big-brand plugins—but the overall takeaway is that it sounds good and greatly simplifies routing.

Unique Points Mentioned (Deduplicated)

  1. Typical dialogue bus uses 4–6 plugins (EQ, bus compression, de-esser, limiter, loudness monitoring).
  2. DXB2 replaces ~5 plugins with one.
  3. DXB is dialogue/vocal custom built (YouTube, films/TV, podcasts, live streaming).
  4. Developer emphasizes efficiency and custom workflow (“one guy” building it).
  5. DXB1 had ugly UI and inefficient DSP; DXB2 rebuilt from the ground up.
  6. Plugin formats mentioned: Pro Tools, AU, VST3 (PC beta referenced as similar code).
  7. UI designed like EQ (high pass, low shelf, 3 mids, air band, low pass).
  8. Mid bands allow notching at problematic frequencies (e.g., around 2k).
  9. Air band includes a frequency-aware de-esser (orange line).
  10. De-esser adjustment/reset via right-click/drag and double-click.
  11. De-esser shown at low CPU (example ~3%).
  12. EQ dynamics are essentially the de-esser within the air band.
  13. Scaled frequency display aims to remain useful across input levels.
  14. Compressor integrated AutoComp.
  15. AutoComp uses RMS + smart threshold + makeup gain to avoid distortion.
  16. Compression can be reduced to 0 (no compression).
  17. Lift control dynamically raises quiet sources to solve “below threshold = no compression.”
  18. Lift can be toggled on/off via right-click.
  19. LUFS monitoring includes input LUFS and output trim.
  20. Output gain mapping targets TV -24 LUFS (creator notes 0 dB => ~-24 LUFS).
  21. Example YouTube workflow uses ~+8 dB output gain to reach -16 to -18 LUFS.
  22. Limiter is zero latency, “quasi-peak,” switchable; off may allow clipping.
  23. Presets supported.
  24. Closing the UI improves efficiency.
  25. Noise reduction placement: creator prefers Voxengo before DXB2.
  26. Voxengo reason: next-gen NR benefits from full-range/noisier input.
  27. Voxengo uses filter bank (not AI reconstruction).
  28. Voxengo does less on already-clean audio; more on noisy sources.
  29. If NR dulls highs, use DXB2 Air/EQ.
  30. Latency measured at ~960 samples (~20 ms) for DXB2 with noise reduction.
  31. Older multi-plugin chain latency up to 8,443 samples.
  32. DXB2 recommended for simplifying dialogue processing.
  33. Mentions a free 2-week trial for Mac and PC.

Speakers’ Views

  • A single main speaker (creator/developer “Tom”) presents all details, demos, comparisons, and recommendations.
  • No other perspectives are clearly included beyond brief quoted examples/audio snippets.

Original video