Prep Production Audio for Post: Alts, Metadata, and Delivery
Executive Summary
The gap between a smooth post-production audio workflow and a painful one almost always traces back to decisions made on set. This guide covers the four pillars of production audio prep: preserving every take (including alts), embedding accurate metadata, maintaining proper gain staging, and assembling a complete delivery package. Each section is grounded in professional practice and references from acknowledged masters of the craft.
Table of Contents
- Don't Delete Alts: The Value of Every Take
---
Don't Delete Alts: The Value of Every Take
One of the most common pitfalls in production sound is the premature judgment and deletion of "unusable" takes or alternate recordings. The mantra in professional sound post-production is clear: never delete alts. Every single take, regardless of perceived quality on set, holds potential value for the sound editor and mixer. What might sound like a compromised take in the field could contain a crucial line reading, a specific emotional nuance, or a piece of clean room tone that becomes invaluable later.
Consider the work of Walter Murch, an editor known for his meticulous approach to sound and picture. In his book In the Blink of an Eye, he discusses the importance of having options, even seemingly imperfect ones, to discover the true rhythm and emotional core of a scene. Deleting an alt on set removes a potential choice for the editor and mixer, limiting their creative palette. The production sound mixer's job is to record everything, not to pre-edit.
This principle extends beyond primary dialogue takes. It includes slates, wild lines (dialogue recorded separately from the main action), and room tone. Room tone, in particular, is often undervalued on set. It is the ambient sound of a location without any dialogue or specific action, recorded for at least 30 to 60 seconds. This seemingly innocuous recording is essential for filling gaps in dialogue edits, smoothing transitions, and creating a consistent sonic bed throughout a scene. Without clean room tone from the exact location and time of day, post-production will struggle to blend dialogue convincingly, often resorting to generic library sounds that can subtly undermine the scene's realism.
MASTER STUDY: No Country for Old Men (Skip Lievsay)
Industry practice mandates archiving full multi-tracks without trimming. This means if a scene was recorded with multiple lavaliers, boom mics, and perhaps a plant mic, all those tracks should be delivered to post, even if only one or two were deemed "primary" on set. Post teams frequently need to reconstruct dynamics or select optimal performances by blending different microphones. For instance, a lavalier might capture a clearer line reading, but the boom mic might offer a more natural, spacious sound. The ability to combine elements from various tracks provides immense flexibility. For practical strategies on mic selection by scenario, see our guide on Lav Strategy by Wardrobe Type.
💡 Pro Tip: When a take is clearly unusable (e.g., mic hit, major siren), log it explicitly in your sound report with a detailed reason, but still record and deliver the track. This clarifies the issue for post and prevents them from wasting time analyzing it, while still preserving the raw data in case a miracle is needed.
Metadata: The Language of Post-Production
Metadata is the invisible backbone of an efficient audio workflow. It is data about data: information embedded within or alongside your audio files that provides critical context for post-production. Without accurate, well-organized metadata, the post-sound team is forced to spend valuable time and resources performing "detective work," trying to decipher what was recorded, by whom, and under what conditions.
Think of it as the sound equivalent of a camera report with detailed lens, f-stop, and lighting information. Just as a DP logs their choices to inform the colorist, a sound mixer logs their choices and observations to inform the sound editor and mixer. This is not just about labeling; it is about embedding intelligence into the audio assets themselves.
Key metadata points include: take numbers, mic positions (e.g., "boom overhead," "lav actor A chest," "plant mic by door"), gain levels, and environmental notes. These notes are crucial. Did a refrigerator hum during the take? Was there a distant dog barking that needs to be addressed? Was the mic off-axis to avoid plosives? Logging these details directly into the sound report, and ideally as embedded notes in the audio files, allows the post-production team to anticipate and address issues proactively. For a deeper breakdown of what post teams actually use from these reports, see Production Sound Reports: What Post Actually Uses and Why.
Embedded metadata also facilitates searchable, automated post workflows. When a sound editor receives files with properly tagged take numbers and mic descriptions, they can quickly sort, filter, and identify specific audio elements. This significantly speeds up the editing process, allowing more time for creative decisions rather than organizational tasks.
💡 Pro Tip: Beyond standard scene and take information, create a consistent system for logging specific "gotchas" in your sound report, such as "airplane overhead," "actor stepped on mic cable," or "loud costume rustle." This helps post prioritize problematic takes and prevents them from wasting time on unfixable audio.
On-Set Gain Staging and Real-Time QC
The quality of the final mix is fundamentally dependent on the quality of the recordings. On-set gain staging and real-time quality control are the first line of defense against irreparable audio issues. Proper gain staging involves setting microphone input levels to capture dialogue clearly without clipping (distortion from levels being too high) or being too quiet (which amplifies noise when boosted in post). A common rule of thumb is to aim for healthy levels that hover around -12 dBFS to -6 dBFS on digital meters, leaving sufficient headroom for unexpected loud moments.
Microphone placement is equally critical. For dialogue, maintaining a consistent distance (often 6 to 12 inches) from the source is ideal. When possible, positioning a boom mic slightly off-axis (not pointed directly at the mouth, but slightly above or to the side) can minimize plosives (P and B sounds that cause bursts of air) and sibilance (harsh S sounds). This technique, combined with a pop filter on the microphone, reduces the need for aggressive de-essing or de-plosive processing in post, which can often make dialogue sound unnatural.
Monitoring on set is non-negotiable. The sound mixer and boom operator must wear high-quality, closed-back headphones to accurately assess the audio being recorded. Closed-back headphones are crucial because they block out ambient set noise, allowing for a clearer perception of microphone pickup, subtle environmental sounds, and potential issues like extraneous noises or clothing rustle. Monitoring in mono on headphones can also help identify phase issues, which occur when multiple microphones capture the same sound at slightly different times, leading to hollowness or cancellation.
Real-time quality control also involves paying attention to the environment. Are there air conditioners running? Refrigerators humming? Distant construction? These sounds, often unnoticed during the chaos of production, can become glaring issues in a quiet movie theater or over headphones. Logging these environmental factors in the sound report is essential. For strategies on evaluating locations before the shoot, see our guide on How to Scout Locations for Sound.
Metering plugins, such as those found in NUGEN VisLM or iZotope Insight 2, are becoming increasingly common on set. These tools provide real-time LUFS and true peak warnings, allowing sound mixers to identify potential loudness or clipping issues as they happen. This proactive approach ensures that the audio handed over to post is technically sound and meets initial broadcast or streaming loudness requirements.
AI-Assisted Noise Reduction: Preserve the Originals
The advent of AI-assisted noise reduction and source separation tools has changed post-production audio workflows significantly, but their effective use hinges on a critical principle: always preserve the original, unprocessed audio. These powerful tools can isolate dialogue from background noise, de-reverb, and even separate speech, music, and ambience into distinct stems from a single audio file. However, they are not a magic bullet, and aggressive processing can introduce artifacts or degrade the natural quality of the sound.
Tools like Adobe Podcast Enhance Speech and DaVinci Resolve Fairlight's built-in noise restoration capabilities offer impressive results. They can clean up noisy dialogue, remove hums, and even de-mix content from uncontrolled environments. This is particularly useful for productions that could not afford pristine recording conditions or for archival footage.
The key is to apply these processes non-destructively. This means performing noise reduction or source separation on copies of the original files, or within a DAW environment that allows for easy A/B comparison with the unprocessed version. Post-production sound designers often use source separation on alternate takes first. They might mute ambience stems entirely for dialogue rebuilds, then layer back selective elements to maintain realism. This granular control allows for fine-tuning the balance between clarity and naturalness.
The "AI artifacts" phenomenon is real. Over-processing can make dialogue sound robotic, phasey, or unnatural. A common mistake is to over-rely on AI without having the original files for comparison, leading to irreversible sonic degradation. Therefore, the production sound team's role in logging background noise and providing clean room tone remains paramount. These elements inform the post-production team on how to apply AI tools, guiding them to use the least aggressive processing necessary for the desired result.
💡 Pro Tip: When recording room tone, capture it not just for the entire location, but also specifically for different microphone positions or camera setups within a scene. A lavalier will often pick up a different quality of room tone than a boom mic, and having specific room tone for each mic can be invaluable for smooth editing.
Production Sound Reports and Deliverables
The production sound report is the central communication document between the set and post-production. It is more than just a list of takes; it is a comprehensive log that provides the context and technical data needed to efficiently process the audio. A well-crafted sound report can save days, if not weeks, of work in post.
Modern industry standards require these reports to be exceptionally detailed, often including timecode-accurate logs, and critically, loudness metadata. This includes LUFS (integrated, short-term, momentary), true peak, LRA (Loudness Range), and DR (Dynamic Range). These metrics are generated by on-set metering plugins like NUGEN VisLM or iZotope Insight 2 and are invaluable for guaranteeing that the final audio will meet distribution specifications. For example, if a report indicates that a take has a true peak violation, the post-mixer knows immediately to address it to prevent clipping during playback on consumer devices.
These reports are typically delivered in Excel or CSV formats, making them easily importable into DAWs for metadata-driven editing. This means a post-mixer can load a scene, see the LUFS data for each clip, and instantly understand its loudness profile without manual analysis. This not only streamlines the workflow but also ensures compliance with platforms that have strict loudness requirements (many streaming services specify -14 LUFS integrated).
The final delivery package to post-production should be a meticulously organized, zipped archive. It must contain all multi-tracks (including all alts, wild lines, and room tone), the comprehensive sound reports (with all metadata), and ideally, synced transcripts for accessibility and text-based editing. The audio files themselves should be industry-standard 24-bit, 48 kHz WAV files for maximum compatibility and quality. For context on how these files connect to the picture editor's workflow, see our Sound Turnover Checklist for Picture Editors.
Before handover, a final validation using metering tools like NUGEN VisLM or iZotope Insight 2 is crucial. This audit confirms that all metrics, especially loudness and true peak, align with the report and meet any specified delivery requirements. Missing room tones, incomplete zips, or reports lacking true peak/DR validation are common mistakes that create significant headaches for the post-production team.
💡 Pro Tip: Include a "reference mix" WAV file (an unmastered, rough mix of the production audio) with your delivery. This provides the post-production team with an immediate understanding of the intended sonic balance and tone for the scene, guiding their mixing decisions.
Common Mistakes
* Deleting "bad" takes: The most common and damaging mistake. What seems unusable on set might contain a critical piece of dialogue or atmosphere that cannot be re-recorded.
* Poor or inconsistent metadata: Lack of clear take numbers, scene names, mic descriptions, or environmental notes creates immense "detective work" for post.
* No room tone: Failing to record sufficient, clean room tone from each unique location and setup makes smooth dialogue editing nearly impossible.
* Improper gain staging: Recording audio too hot (clipping) or too quiet (requiring excessive boosting, which raises the noise floor) creates irreversible quality issues.
* Inadequate monitoring: Not using closed-back headphones or failing to listen critically on set means problems are missed until post, by which point it is too late.
* Aggressive on-set processing: Applying noise reduction or EQ destructively on set, rather than letting post handle it with more sophisticated tools and context.
* Disorganized file delivery: Delivering unzipped, unlabeled, or incomplete audio files and reports complicates the entire post-production ingest process.
* Ignoring loudness metadata: Not generating or including LUFS, true peak, and LRA data in the sound reports means post has to re-analyze everything, slowing down compliance checks.
Interface and Handoff Notes
What You Receive (Upstream Inputs)
* Script/Shot List: Provides context for recording and logging. * Picture Editor's Cut (Reference Video): For syncing and review. * Production Schedule: To anticipate recording needs.
What You Deliver (Downstream Outputs)
* Multi-track WAV Files: Organized by scene/take, pristine, unprocessed originals. * Comprehensive Production Sound Reports: Detailed logs with timecode, mic info, notes, and loudness metadata (LUFS, True Peak, LRA). * Room Tone Library: Clearly labeled and categorized room tone for all locations/setups. * Wild Lines/ADR Tracks: Any separately recorded dialogue or sound elements. * Zipped Delivery Package: All files and reports securely compressed for transport.
Top 3 Failure Modes
1. Missing or Corrupted Takes: Due to premature deletion of alts or data loss, forcing re-recording or creative workarounds in post.
Practical Templates
| Item | Format | Status |
|---|---|---|
| Multi-track WAV files (all takes including alts) | 24-bit / 48 kHz BWF | ☐ |
| Room tone per location/setup | 24-bit / 48 kHz WAV, min 60 sec | ☐ |
| Wild lines / ADR | 24-bit / 48 kHz WAV | ☐ |
| Production sound report | Excel / CSV | ☐ |
| Loudness metadata (LUFS, True Peak, LRA) | Embedded or CSV column | ☐ |
| Mic position notes per take | Sound report column | ☐ |
| Environmental issue log | Sound report notes | ☐ |
| Reference mix | WAV stereo | ☐ |
| Synced transcripts (if available) | SRT / TXT | ☐ |
| Final zip integrity check | Checksum (MD5/SHA) | ☐ |
| Field | Example Entry | Notes |
|---|---|---|
| Scene | 14A | Match camera report |
| Take | 3 | Include circle takes |
| Timecode In | 01:14:22:08 | From recorder TC |
| Track 1 | Boom overhead | Mic model + position |
| Track 2 | Lav Actor A (chest) | Mic model + mount type |
| Track 3 | Lav Actor B (hair) | Mic model + mount type |
| Track 4 | Plant mic (door frame) | Describe placement |
| Peak Level | -8 dBFS | Highest peak on any track |
| LUFS (Integrated) | -22.4 LUFS | Per take measurement |
| True Peak | -2.1 dBTP | Flag if above -1 dBTP |
| Environmental Notes | AC hum, distant traffic | Log all anomalies |
| Mixer Notes | Actor B whispered, low level | Context for post |
Browse This Cluster
---
© 2026 BlockReel DAO. All rights reserved. Licensed under CC BY-NC-ND 4.0 • No AI Training.