Check-list de Montage Podcast : Pauses, Bruits, Clics, Respirations et Écoute Finale

When to Use This Checklist (and When to Skip Steps)

This checklist is engineered for spoken-word audio creators, podcast producers, and audio engineers who need a rigorous, repeatable quality-control protocol.

Boundary Conditions: Unscripted vs Scripted Content

• Unscripted Multi-Speaker Podcasts: Execute all 6 phases. Pay strict attention to phase alignment between microphones, remote VoIP latency trimming, and multi-track balance.
• Solo Scripted Voiceover & Audiobooks: Skip multi-track bleed checks. Tighten silence cushions (0.15s – 0.25s) and adhere to strict ACX/Audible loudness limits (-18 to -23 LUFS with -3.0 dBTP peak ceiling and noise floor below -60 dBFS).
• Live Broadcasts / Quick Social Clips: Focus on Phase 3 (high-pass filter) and Phase 5 (target loudness normalization).

The Non-Negotiables: Loudness Safety and Pacing Integrity

Regardless of production format, two standards are mandatory:

1. True-Peak Safety: True Peak must never exceed -1.0 dBTP (or -2.0 dBTP for MP3 files) to prevent inter-sample distortion when streaming platforms compress your master.
2. Preserved Speech Cushions: Pauses must never be cut to zero duration. Retaining 0.32 seconds prevents breathless, robotic delivery.

The Complete 6-Phase Podcast Quality Assurance Checklist

Phase 1: Pre-Flight Ingest & Multi-Track Alignment

• [ ] Duplicate Raw Masters: Save untouched copies of all raw recording tracks in an offline backup folder.
• [ ] Inspect Ingest Gain Staging: Verify raw peaks sit comfortably between -18 dBFS and -12 dBFS with zero digital clipping.
• [ ] Synchronize Multi-Track Claps: Align separate guest tracks to a common sync marker (visual clap or tone) down to the millisecond.
• [ ] Check Acoustic Bleed: Mute host mic during long guest answers to prevent room reverb phase cancellation.

Phase 2: Structural Editing & Silence Auditing

• [ ] Run Automated Silence Triage: Detect dead-air gaps exceeding 0.80s at -38 dB sensitivity.
• [ ] Apply 0.32s Retained Cushion: Preserve natural breath beats and phoneme decay at all splice points.
• [ ] Prune False Starts and Retakes: Identify verbal restart phrases ("let me start over", "take two") and prune superseded takes.
• [ ] Enable Subtitle Sentence Protection: Shield contemplative mid-sentence pauses inside active dialogue lines from deletion.

Phase 3: Surgical Acoustic Cleanup (Rumble, Hum, Clicks)

• [ ] Apply 80Hz High-Pass Filter: Strip handling noise, desk thumps, and HVAC low-frequency rumble with a 12dB/octave Butterworth curve.
• [ ] Engage 50Hz / 60Hz Notch Filters: Isolate and attenuate electrical mains grounding hum.
• [ ] Run Transient De-Clicker: Scan for 2kHz–10kHz saliva pops and repair with cubic spline interpolation.
• [ ] Check Plosive Thumps: Ensure hard "p" and "b" sounds do not cause low-frequency transducer overload.

Phase 4: Dynamic Leveling, Breath Softening & Room Tone

• [ ] Soften Heavy Inhalations: Attenuate loud gasps by -6 dB to -12 dB; do not cut 100% of breath sounds.
• [ ] Synthesize Ambient Room Tone: Fill trimmed gaps with looped room acoustics using Overlap-Add (OLA) crossfading to eliminate vacuum blackouts.
• [ ] Apply Broadcast Vocal EQ: Add +1.5 dB body at 200Hz, trim -1.5 dB boxiness at 500Hz, and add +2.0 dB clarity at 3.5kHz.
• [ ] Verify Dynamic Range (LRA): Ensure Loudness Range sits comfortably between 4.0 LU and 8.0 LU.

Phase 5: Master Loudness & True-Peak Verification

• [ ] Set Master Loudness Target:
  • Stereo Episodes: -16.0 LUFS Integrated (±1.0 LUFS).
  • Mono Episodes: -19.0 LUFS Integrated (±1.0 LUFS).
• [ ] Engage True-Peak Limiter: Set hard ceiling to -1.0 dBTP (or -2.0 dBTP for MP3 distribution).
• [ ] Verify Metadata Tags: Embed episode title, podcast name, artwork, and chapter markers into final export.

Phase 6: Multi-Environment Listening & Mono QC

• [ ] Mono Fold-Down Test: Sum the stereo mix to mono and listen for phase comb filtering or disappearing voices.
• [ ] Single-Earbud Commute Test: Audition the track with a single headphone in a mildly noisy environment.
• [ ] Smartphone Speaker Test: Play dialogue through a mobile phone speaker at 50% volume to verify vocal intelligibility.

Technical Specifications & Tolerance Limits

Keep these exact technical targets handy during mastering in the Podcast Mixing Studio:

Step-by-Step QA Workflow in AutoSilenceCut

1. Load Master Audio: Drag your mixed WAV file into Podcast Mixing Studio.
2. Run Automatic Delivery Diagnostic: The engine instantly measures initial Integrated LUFS, True Peak, Noise Floor (dBFS), and Loudness Range.
3. Activate Rack Filters: Toggle on 80Hz High-Pass, Mains Notch, Vocal Presence EQ, and Room Tone Matcher.
4. Audit Pacing Cuts: Review the pause timeline. Use [Hear Before] to ensure no conversational beats were truncated.
5. Export Broadcast Master: Click Export 16-Bit WAV (-16 LUFS) for a broadcast-ready file with embedded ITU-R BS.1770 compliance.

Cas concret : Weekly Interview Episode

A weekly technology podcast recorded a 38-minute episode with a remote guest over Zoom.

• Pre-Flight Findings: Raw audio had -26 LUFS loudness, a 60Hz mains hum from the guest's charger, and 4 minutes 20 seconds of dead-air latency.
• QA Execution: 60Hz notch filter engaged; silence scanner trimmed 3 minutes 50 seconds of remote lag while preserving 0.32s cushions; mouth de-clicker resolved 64 saliva pops.
• Mastering Output: Output normalized from -26.0 LUFS to exactly -16.1 LUFS Integrated with True Peak capped at -1.0 dBTP. Final runtime: 34 minutes 10 seconds.

Common Pitfalls in Podcast Quality Assurance

• Pitfall 1: Normalizing to 0 dBFS Peak Instead of -16 LUFS: Peak normalization only looks at the single loudest sound (like a loud laugh), leaving average speech quiet and inaudible.
• Pitfall 2: Forgetting to Check for Stereo Phase Cancellation: Two mics picking up the same speaker will cancel out when summed to mono unless checked and phase-aligned.
• Pitfall 3: Over-Cleaning Room Tone: Completely muting room tone creates jarring acoustic dropouts every time the speaker stops talking.
• Pitfall 4: Listening Only at High Volume: Human ears perceive frequency balance differently at high volumes (Fletcher-Munson effect). Always do your final listening pass at moderate-to-low conversational volume.