Skip to content

Dialogue & Podcast Pipeline

Mastering tuned for the human voice, not a snare drum.

Dialogue-first loudness, breath and room control, and platform-compliant masters that hit Apple Podcasts, Spotify, and every RSS listener at the exact LUFS, true-peak, and noise floor your host mix was always meant to meet.

Start a Free Master Hear a sample

No credit card. 3 free masters per month. Renders in 47 seconds.

By the numbers — podcast mastering, ringed

Specs a podcast engineer screenshots, not adjectives.

  • Dialogue LUFS target

    -19 LUFS

    Pasted into every Apple, Spotify, and RSS-compliant render — matches the Apple Podcasts v4.2 reference.

  • Integrated loudness tolerance

    ±0.21 LU

    Measured episode-to-episode across a 12-episode season, vs the hand-mastered benchmark of ±0.8 LU.

  • True-peak ceiling

    -1.0 dBTP

    Prevents inter-sample clipping in browsers, mobile carriers, and car Bluetooth playback chains.

  • Episode turnaround

    47 sec

    Median single-episode master on the cloud render farm. A 12-episode season returns in 6 min 14 sec.

Targets verified against Apple Podcasts Connect v4.2, Spotify master ingest, AES TD1008 (streaming loudness recommendation), and EBU R128 s1.

Inside the Dialogue & Podcast pipeline

Four moves a good podcast engineer makes on every voice file — automated, in order, in one pass.

Every master runs through the same four-stage model that's been trained on 18.3 million commercially released, professionally mastered tracks and re-tuned on a 412,000-episode dialogue corpus. No slider shopping, no broken-by-default presets.

  1. 01

    Voice clarity  EQ

    Adaptive de-essing, proximity-band control on the 200–400 Hz muddiness that plagues USB and lavalier mics, and presence lift tuned to the 3–5 kHz intelligibility range. The model picks the curve based on the actual spectral shape of the host, not a static "podcast" preset.

    • De-esser ratio: model-driven, 3:1 to 6:1
    • Proximity cut: 0 to -4.5 dB, learned
    • Presence target: 3.6 kHz reference
  2. 02

    Breath & room control

    Soft-noise gating between phrases and a dynamic room-reverb suppressor tuned for untreated home studios. The model distinguishes "host pause for breath" from "noise floor dip" before it touches either, so the edit never sounds pumped or gated.

    • Noise-floor target: -67 dBFS RMS
    • Breath smoothing: 120 ms onset / 240 ms release
    • Room-tail reduction: 12–18 dB, learned
  3. 03

    Platform loudness compliance

    One render, four delivery targets. AudioMagus applies the LUFS, true-peak, and noise-floor spec that Apple Podcasts, Spotify, YouTube (broadcast), and generic RSS hosts each demand — and exports each as a labelled stem so your CDN can swap on the fly.

    • Apple Podcasts: -19 LUFS, -1 dBTP
    • Spotify master ingest: -14 LUFS, -1 dBTP
    • RSS / generic: -16 LUFS, -1 dBTP
    • YouTube broadcast: -14 LUFS, -1 dBTP
  4. 04

    Season consistency

    Submit a 12-episode season as a single batch and the model locks the same loudness, EQ character, and noise floor across all of them — even when the files were recorded months apart, on different mics, in different rooms. This is the stage hand-mastered series almost never get right.

    • Cross-episode LU spread: ≤ 0.21 LU
    • Cross-episode noise-floor spread: ≤ 0.9 dB
    • Master chain reference dither: POW-R 1

Master your next episode in under a minute

Drop in your host mix. Hear what platform-compliant dialogue mastering sounds like — without booking an engineer.

3 free masters every month on the free tier. No credit card. No "contact us for podcast pricing." Render a single episode first; bring the whole season when you're ready.

Start a Free Master Read the engine spec
  • WAV, MP3, FLAC, AAC up to 1.2 GB per episode
  • Stems exported per platform (Apple / Spotify / RSS / YouTube)
  • SOC 2 Type II — your unreleased material stays encrypted