Audio Delivery Standards for Film and Television: A Complete Format Guide

The Delivery Rejection Nobody Saw Coming

A sound mixer finishes the mix for a short documentary. It sounds great on the studio monitors -- full, dynamic, punchy. The film submits to a major broadcast festival. Three weeks later, a rejection arrives. The audio is technically non-compliant: the integrated loudness measures -18 LUFS, the broadcaster requires -23 LUFS per EBU R128, and the true peak on two dialogue moments exceeds -1 dBTP. The film is invited to resubmit with a corrected mix, but the sound mixer's invoice has already been paid and they've moved on to another project.

A 90-minute re-mix negotiation later, the re-mastered mix ships two days before the festival deadline. It makes the cut -- but the near-miss cost $800 in additional sound work and two weeks of avoidable stress.

Every platform, broadcaster, and festival that accepts deliverables has a technical specification. Most are publicly available. Almost nobody reads them until a rejection forces the issue.

This post compiles the critical audio specifications for the platforms and delivery channels that matter most to indie filmmakers, so you can hand them to your sound mixer before the mix begins, not after it fails.

Audio specifications referenced in this guide are drawn from Netflix's Open Content Delivery Specifications, the EBU R128 loudness standard (European Broadcasting Union), the SMPTE ST 2095-1 standard for theatrical audio, and publicly available delivery requirements from Amazon Prime Video, Apple TV+, and major US broadcasters.

How Audio Delivery Specs Actually Work

Three technical parameters govern audio compliance across every platform: integrated loudness, true peak level, and channel configuration. Understanding what each parameter means -- and why it matters to the platform -- makes the spec table that follows more than a reference list.

Integrated loudness is measured in LUFS (Loudness Units Full Scale) and represents the average perceived loudness of the entire program from start to finish. The measurement algorithm, defined in EBU R128 and ITU-R BS.1770, weights frequency bands to approximate human hearing, meaning loud high-frequency content registers differently than an equivalent SPL at low frequency. Platforms normalize to a target loudness value so that a quiet documentary and an action film play back at comparable perceived levels without the viewer needing to adjust volume between titles. The target for broadcast and most streaming is -23 LUFS (EBU) or -24 LUFS (ATSC A/85, used by US broadcasters). Netflix uses -27 LUFS for theatrical content.

True peak level is the maximum instantaneous sample value in the audio, measured in dBTP (decibels True Peak). It differs from peak metering because it accounts for inter-sample peaks -- momentary excursions that occur between samples during digital-to-analog conversion and can cause clipping even when the peak meter reads below 0 dBFS. The universal limit across broadcast and streaming platforms is -1 dBTP. Film theatrical delivery allows up to -2 dBTP for the LFE (low-frequency effects) channel.

Channel configuration defines how many audio tracks are delivered and what signal goes to each. Stereo (2.0) is the minimum for streaming. 5.1 surround (Left, Right, Center, LFE, Left Surround, Right Surround) is required for theatrical DCP delivery and for premium streaming tiers on Netflix, Amazon, and Apple TV+. Dolby Atmos delivery adds object-based audio metadata on top of the 7.1 bed and is increasingly requested by premium streaming platforms. Most broadcasters require a discrete stereo mix plus a 5.1 mix delivered as separate files.

Streaming and Broadcast Audio Delivery Specifications

The most significant variance in this table is between YouTube (-14 LUFS) and Netflix theatrical (-27 LUFS). A mix mastered for YouTube sounds excessively loud when normalized down to -27 LUFS on Netflix -- the dynamic range compression that was invisible at -14 LUFS becomes audible at lower playback levels. This is why dialogue-heavy content sounds noticeably less punchy on premium streaming services than on YouTube: the platforms are playing back at genuinely different loudness levels, and mixes need to be structured with target platform in mind.

Three Real-World Delivery Scenarios

Example 1: Short Film Festival Circuit Submission

A 22-minute documentary short heading to festival submission. Target festivals include Sundance, SXSW, and a range of smaller documentary festivals. No broadcast deal exists yet.

Specification used: -24 LUFS integrated, -1 dBTP true peak, 48kHz / 24-bit stereo plus a separate 5.1 mix delivered as a 6-channel interleaved WAV (in LCR / LFE / LS / RS channel order).

Outcome: The mixer measured the final stereo mix at -21.4 LUFS integrated with a true peak of -0.8 dBTP. A 2-stop loudness reduction (technically: -2.6 LU reduction) using Izotope RX's match loudness processor brought the mix to -24.0 LUFS with true peak remaining compliant. The dynamic character of the mix was preserved because the reduction was uniform, not limited.

Key lesson: Measuring loudness at the end of the mix and adjusting with a true-peak-aware limiter is standard professional practice. Building the mix toward -24 LUFS from the start avoids the adjustment step entirely.

Example 2: Feature Film, Amazon Prime Video Delivery

A 94-minute narrative feature licensed to Amazon Prime Video. The delivery specification requires a -24 LUFS stereo mix, a -24 LUFS 5.1 mix, and an Dolby Atmos ADM BWF file for Atmos-capable devices.

Specification used: Amazon Prime Video Delivery Specification v2.2 -- -24 LUFS integrated (±1 LU tolerance), -2 dBTP true peak, 48kHz / 24-bit for all audio files. Atmos ADM BWF at 48kHz / 24-bit.

Outcome: The re-recording mixer delivered four audio files: a stereo LtRt (Lt/Rt encoded for Dolby Surround downmix compatibility), a discrete stereo, a 5.1 interleaved WAV, and the Atmos ADM. The Atmos mix required a separate deliverables session of approximately 12 hours beyond the 5.1 mix to map dialogue, music, and effects to appropriate bed and object channels.

Key lesson: Atmos delivery adds roughly 10-15% to mixing costs. If Atmos delivery is contractually required, it must be in the mixing budget from the start -- it cannot be bolted on at the end of a standard 5.1 mix session without extra time.

Example 3: Corporate Brand Film, YouTube and LinkedIn

A 6-minute brand documentary for a corporate client. Primary delivery is YouTube and the client's LinkedIn page. No theatrical or broadcast distribution.

Specification used: -14 LUFS integrated (YouTube normalization target), -1 dBTP, 48kHz / 16-bit stereo WAV.

Outcome: The sound designer mixed at -14 LUFS from the start, which allowed a louder, more impactful mix compared to a broadcast spec -- appropriate for the corporate context where the client wanted the brand film to feel energetic on social media. The 16-bit delivery at 48kHz met YouTube's minimum spec and reduced file size for the client's internal asset library.

Key lesson: Mixing to the delivery platform's target loudness -- rather than applying a generic loudness correction at the end -- produces a better result because dynamic range decisions made during the mix are appropriate for the playback context.

Audio Delivery Preparation: A Step-by-Step Checklist

Step 1: Obtain the platform delivery specification before the mix begins. Netflix, Amazon, Apple TV+, and most broadcasters publish their tech specs on their production portals. Download the current version -- specs update regularly and what was compliant in 2023 may have stricter tolerances in 2026.

Step 2: Confirm channel configuration requirements with the distributor or festival. A festival requiring a 5.1 mix when you've only mixed in stereo means a remixing session, not a channel map. Confirm before the final mix -- not after picture lock.

Step 3: Mix with a loudness meter active and visible throughout the session. Real-time LUFS metering is built into DaVinci Resolve's Fairlight page, Adobe Audition, Pro Tools, and Nuendo. Set your integrated loudness target as a reference marker. Don't wait until the end of the mix to check compliance -- mix toward the target number the entire session.

Step 4: Use the [LUFS Loudness Calculator](/tools/lufs-loudness) to check your integrated measurement against the delivery target before final export. The calculator allows you to input your measured LUFS value and determine the exact gain adjustment needed to reach compliance.

Step 5: Apply true peak limiting as the final processing step. After all mix processing, insert a true peak limiter (iZotope RX Loudness Control, Waves WLM Plus, or Massey L2007 configured for true peak mode) set to the platform's maximum -- typically -1 dBTP or -2 dBTP. Check that the limiter is operating in inter-sample peak mode, not standard peak mode. Standard peak limiters will not catch all true peak violations.

Step 6: Export at 48kHz / 24-bit unless the platform specifically requires 44.1kHz. 48kHz is the professional standard for all video audio. 44.1kHz is the legacy audio CD standard and appropriate only for music-only deliverables to services like Spotify and Apple Music. Delivering video audio at 44.1kHz to a platform requiring 48kHz will result in a compliance failure or a sample rate conversion that degrades audio quality.

Step 7: Verify file format and naming conventions. Netflix requires MXF audio files with specific naming conventions. Most festivals accept stereo WAV or AIFF. DCP delivery requires MXF-wrapped audio compliant with SMPTE ST 429-2. Confirm file format separately from loudness -- both must be correct for a submission to be accepted.

Pro Tips and Common Mistakes

Pro Tip: For festival submissions where the delivery spec is not published, use -24 LUFS and -1 dBTP as the safe default. This falls within EBU R128 broadcast spec and is compliant with virtually every professional exhibition and streaming context. Only deviate from this default when a platform's published specification explicitly requires a different target.

Pro Tip: When delivering to YouTube, don't over-compress your mix in an attempt to hit -14 LUFS with maximum apparent loudness. YouTube's loudness normalization algorithm measures integrated loudness over the entire program. A mix that's heavily limited to -14 LUFS will still be brought down to -14 LUFS in playback -- but the limiting will be audible. A mix with real dynamics that averages -14 LUFS sounds better in the final playback than an over-limited one at the same integrated level.

Pro Tip: Deliver a stereo mixdown along with every 5.1 or Atmos deliverable. Even platforms that support surround sound serve a significant portion of their audience on devices that only play stereo. A 5.1 downmix algorithm can produce phase issues on stereo playback. A purpose-mixed stereo file sounds better and prevents listening fatigue caused by folded surround channels.

Common Mistake: Measuring loudness with a peak meter and assuming LUFS compliance. A peak meter shows the highest instantaneous sample value -- it tells you nothing about perceived loudness or integrated LUFS. A scene with quiet dialogue and one very loud sound effect can measure -6 dBFS on a peak meter while integrating at -32 LUFS -- wildly non-compliant in the opposite direction from what the peak number suggests. Always use a dedicated LUFS meter.

Common Mistake: Delivering a stereo file as a "5.1 mix" by duplicating the stereo signal across all six channels. This is immediately identifiable in compliance QC and results in delivery rejection. A stereo-to-5.1 upmix performed by a processor like Nuendo's Upmix plugin or iZotope RX is technically more defensible -- but most platforms and broadcasters require a native 5.1 mix, meaning one that was either recorded or mixed in surround from the start.

Frequently Asked Questions

What's the difference between integrated loudness and short-term loudness?

Integrated loudness measures the average perceived loudness of the entire program from start to finish -- the number that must hit the platform's LUFS target. Short-term loudness measures the same metric over a sliding 3-second window, providing a real-time display of loudness as the mix evolves. Momentary loudness measures over a 400-millisecond window, showing loud transients. Delivery specifications almost always specify integrated loudness. Short-term and momentary loudness are mixing tools, not delivery compliance targets.

Why does Netflix use -27 LUFS when most broadcasters use -23 or -24 LUFS?

Netflix calibrates its theatrical content delivery at a lower loudness target to preserve the dynamic range of cinematic mixes -- the difference between a whispered line and an explosion is intentional and should be maintained in home playback. The -27 LUFS target is aligned with theatrical monitoring levels in a calibrated room (85 dB SPL reference with the meter at -20 dBFS). At -27 LUFS, a properly mixed film retains its full cinematic dynamic range without the compression required to reach -23 LUFS. Netflix's spec applies to content delivered for their premium tier; short-form content on the platform may normalize differently.

How do I handle audio delivery when my mix has very wide dynamics -- like a documentary with whispered interviews and loud crowd scenes?

Wide-dynamic mixes that comply with integrated loudness targets are technically valid -- the integrated target averages across the entire duration. A documentary with quiet interviews at -40 LUFS and one loud crowd scene at -12 LUFS can still integrate to -24 LUFS if the quiet material dominates the runtime. The challenge is true peak compliance: the loud scenes must still stay below -1 dBTP. Apply true peak limiting only to those loud peaks using dynamic range-preserving limiting rather than broadband compression, which would reduce loudness across the quiet sections too.

What audio format does a DCP require?

A Digital Cinema Package (DCP) per SMPTE ST 429-2 requires audio packaged as MXF-wrapped, 24-bit PCM at either 48kHz or 96kHz. 5.1 channel configuration is standard (Left, Right, Center, LFE, Left Surround, Right Surround). Channel ordering in the MXF must follow the SMPTE specification -- getting the channel order wrong is one of the most common DCP audio errors. DCP audio is not loudness-normalized by the standard; the exhibition space's calibrated room handles playback level. However, most DCP service providers request a mix at approximately -20 dBFS average to ensure appropriate headroom on theatrical playback systems.

Can I use the camera's built-in audio as a final deliverable?

No, except in rare circumstances for micro-budget online-only content. Camera audio recorded to H.264 video files is typically compressed, sample-rate limited, and lacks the noise floor and dynamic range of a dedicated field recorder. The standard minimum for professional delivery is a 48kHz / 24-bit WAV file recorded to a Sound Devices, Zoom F-series, or Tascam DR recorder. If camera audio is the only recording available, it can be cleaned and delivered -- but plan for additional time in the audio post session to correct the noise floor, EQ, and dynamics.

Related Tools

The LUFS Loudness Calculator lets you enter your measured integrated loudness and target platform to calculate the exact gain adjustment needed for compliance -- no manual arithmetic required. For understanding how audio file format choices affect storage and delivery, the Audio Bitrate Storage Calculator computes file size for any combination of sample rate, bit depth, channel count, and duration. The post on LUFS, dBFS, and loudness normalization covers the underlying metering concepts in more depth if the spec table above raises questions about why the numbers are what they are. If your delivery also involves closed captions, closed captions and subtitles for film delivery covers format requirements across the same platforms covered here.

Conclusion

Delivery compliance is not a post-production afterthought -- it's a specification that should be in your sound mixer's brief before the mix begins. The LUFS target determines mix dynamics. The channel configuration requirement determines the session structure. Neither can be corrected cheaply at the end of a locked mix.

This guide covers audio-only delivery parameters. Complete deliverable packages -- including video codec, subtitle file formats, and metadata requirements -- vary by platform and should be obtained directly from the platform's technical delivery specification documents.

Have you ever had an audio deliverable rejected for compliance reasons? What was the violation, and at what stage of post did you discover it?