Netflix Sound Mix Specifications & Best Practices v1.5

Average loudness must be -27 LKFS +/- 2 LU dialog-gated. Peaks must not exceed -2db True Peak. Audio should be measured over the full program according to ITU-R BS.1770-1 guidelines.

Discrete audio is required, except in the following circumstances:

When muxed audio is inherent to the source, such as audio in an IMF, Quicktime or Atmos® BWAV ADM container.
Interleaved Secondary Audio .wav deliveries.

Sample Rate and Bit Depth:

48k/24-bit for Original Language Mix or M&E Mix - applies to Stems & Mix Masters.

48k/16-bit or 24-bit for Secondary Language or Audio Description.

Lossy or lossless audio compression is never allowed. Audio must be standard or RF64 discrete LPCM .wav or .bwav files.

Audio must be recorded in sync to picture at native project frame rate. All audio deliverables should match the camera and picture post production frame rate. All audio must sync to the final IMF/Prores picture delivery. Alternate/Dubbed Language and Audio Description Mixes must not contain an additional leader. Printmasters, M&Es and stems may include standard 8 second Academy leader and 2-pops.

Content Hub deliveries may be discreet audio files alone, within folders or within Pro Tools session folders also containing sessions. Where project files other than Pro Tools are delivered, AAFs/OMFs should also be included, or all channels must be rendered to continuous PCM audio.

For all titles, 5.1 audio is required and 2.0 is optional.

Mono audio is acceptable if the program’s original source is mono and no stereo and/or 5.1 mix exists. Mono audio must be duplicated on channels 1 & 2 and delivered as 2-channel.

Deliver all Near Field audio assets as a longplay (not reels).

2.1 Near Field Atmos® Mix - Mastering and Archive - Original Language Version

The Dolby Atmos® Original Language Home Mix must meet the following requirements:

At a minimum, must be mixed in a 7.1.4 room.
Dolby Certification is not required for Home Atmos® mix rooms.
Atmos® delivered in IMF (IAB) supports 23.976, 24, 25 and 30 frames per second.
This is a near field mix. The most common monitoring levels for near field are 79db or 82db.
Must conform to Netflix loudness spec (-27db LKFS +/- 2 LU 1770-1 dialog-gated).
Peaks should generally not exceed -2dBFS true-peak. To achieve this, we recommend setting true-peak limiters on all beds and objects at -2.3 or lower. Loudness and peaks should be measured via a 5.1 rerender.
For Atmos® mixes, delivering 5.1 is required and 2.0 audio is optional. The 5.1 and 2.0 re-render outputs must conform to Netflix loudness specification. 5.1 and 2.0 must conform to the Netflix peak specification only if they are delivered for streaming. We suggest achieving this by limiting the output from the renderer, rather than lowering beds and objects.
- Atmos® - Delivered to picture finishing facility for IMF creation and streaming.
- 5.1 - Delivered to Content Hub for dubbing reference. Optional to deliver for streaming.
- 2.0 - Optional.
All beds and objects are encoded and will always be audible. A printmaster may utilize a composite of all bed material OR multiple beds, but not both.
For all series/doc/unscripted/comedy content, beds and objects must be separated by Dialog, Music and Effects and must be labeled only as “Dialog”, “Music”, “Effects” or “Narration.” For film content, DME separation should be maintained whenever possible.
If an 85 db reference theatrical mix is created, two complete sets of deliverables are required. One for theatrical, one for near field.
Leader and sync pop are preferred but not required. The IMF vendor will sync Atmos® in the IMF.
Final Home Theater Print Master:
- Dolby Atmos® BWAV ADM File.
  - All bed channel configurations are acceptable.
  - 48 kHz sample rate, 24-bit.

2.2 Near Field 5.1 Surround Mix - Mastering and Archive - Original Language Version

Use 79dB spl or 82dB spl as your standard reference level for mixing.
Meet a -27 LKFS (+/- 2 LU) dialogue loudness using ITU-R BS.1770-1 measured over entire program.
Do not exceed +18db (-2 dBFS) maximum level (true-peak) over reference of -20 dBFS, achieved by peak limiting and not lowering the mix level.
For Original Version: Provide 5.1 Dialog, Music and Effects stems that equal the 5.1 mix when combined.

2.3 Near Field 2.0 Stereo Mix - Mastering and Archive - Original Language Version

Use 79dB spl or 82dB spl as your standard reference level for mixing.
Provide a Lo/Ro or LT/RT mix with -27 LKFS (+/- 2 LU) dialogue loudness using ITU-R BS.1770-1 measured over entire program.
- Lo/Ro mix is preferred.
Lo/Ro or LT/RT mix to be mono compatible.
Do not exceed +18db (-2 dBFS) maximum level (true-peak) over reference of -20 dBFS, achieved by peak limiting and not lowering the mix level.
When down-mixing from 5.1 to create Lo/Ro or LT/RT 2.0 mix:
- Lower Center channel content by -3db.
- Center channel content should be included in both the LEFT and RIGHT CHANNELS.
- Lower Surround channel content by a minimum of -3db.
- Fold Left and Right Surround content into the corresponding LEFT and RIGHT CHANNELS.
- Optionally include the Low Frequency Effects (LFE) channel between -8db and -12db.
  - Fold Low Frequency Effects (LFE) into both the LEFT and RIGHT CHANNELS.
  - Low Pass Filter LFE at 200Hz or lower.
All adjustments must be made to the individual 5.1 channels prior to folding down.

Note: The resulting 2.0 mix must be checked prior to delivery to ensure it is free of any audio artifacts and if necessary, as determined by Netflix, further adjustments shall be made to prevent them. Any 2.0 delivery must meet the specification prescribed in this section.

2.4 Near Field Atmos® Mix - Mastering and Archive - Dubbed Languages & Audio Description

The Dolby Atmos® Dubbed Language Home Mix must meet the following requirements.

At a minimum, must be mixed in a 7.1.4 room.
Dolby Certification is not required for Home Atmos® mix rooms.
This is a near field mix. The most common monitoring levels for near field are 79db or 82db.
Must conform to Netflix loudness spec (-27db LKFS +/- 2 LU 1770-1 dialogue loudness).
Peaks should generally not exceed -2dbfs True Peak. To achieve this, we recommend setting True Peak limiters on all beds and objects at -2.3 or lower. Loudness and peaks should be measured via a 5.1 rerender.
For native Atmos® mixes, the 5.1 and 2.0 rerenders must conform to Netflix loudness and Peak specification.
All beds and objects are encoded and will always be audible. A printmaster may utilize a composite of all bed material OR multiple beds, but not both.
For all series/doc/unscripted/comedy content, beds and objects must be separated by Dialog, Music and Effects. For film content, DME separation should be maintained whenever possible.
If a 85 db reference theatrical mix is created, two complete sets of deliverables are required. One for theatrical, one for near field.
Dubbed Language Atmos® files shall match final IMF picture length exactly as long-play and not separate reels. Leader and sync pop shall be removed.
FFOA must be left at default - leave “Add FFOA” box unchecked. For pre-3.0 renderer software, leave red dashes in FFOA value.
Final Home Theater Print Master:
- Dolby Atmos® BWAV ADM File.
  - All bed channel configurations are acceptable.
  - 48 kHz, 24 bit.

2.5 Near Field 5.1 Surround Mix - Mastering and Archive - Dubbed Languages and Audio Description

Use 79dB spl or 82dB spl as your standard reference level for mixing.
Meet a -27 LKFS (+/- 2 LU) dialog-gated using ITU-R BS.1770-1 measured over entire program.
Do not exceed +18db (-2 dBFS) maximum level (true-peak) over reference of -20 dBFS, achieved by peak limiting and not lowering the mix level.
For Original Version: Provide 5.1 Dialog, Music and Effects stems that equal the 5.1 mix when combined.
For Dubbed Version: Provide 5.1 Dialog stem that equals the 5.1 mix when combined with M&E.

2.6 Near Field 2.0 Stereo Mix - Mastering and Archive - Dubbed Languages and Audio Description

Use 79dB spl or 82dB spl as your standard reference level for mixing.
Provide a separate Lo/Ro or LT/RT mix with -27 LKFS (+/- 2 LU) dialogue loudness using ITU-R BS.1770-1 measured over the entire program.
- Lo/Ro mix is preferred.
Lo/Ro or LT/RT mix to be mono compatible.
Do not exceed +18db (-2 dBFS) maximum level (true-peak) over reference of -20 dBFS, achieved by peak limiting and not lowering the mix level.
When down-mixing from 5.1 to create Lo/Ro or LT/RT 2.0 mix:
Lower Center channel content by -3db.
Center channel content should be included in both the LEFT and RIGHT CHANNELS.
Lower Surround channel content by a minimum of -3db.
Fold Left and Right Surround content into the corresponding LEFT and RIGHT CHANNELS.
Optionally include the Low Frequency Effects (LFE) channel between -8db and -12db.
- Fold Low Frequency Effects (LFE) into both the LEFT and RIGHT CHANNELS.
- Low Pass Filter LFE at 200Hz or lower.
All adjustments must be made to the individual 5.1 channels prior to folding down.

2.7 Music & Effects (M&E) Mix

2.7.1 Near Field Atmos® M&E Mix

The Dolby Atmos® Home M&E Mix must meet the following requirements:

At a minimum, must be mixed in a 7.1.4 room.
Dolby Certification is not required for Home Atmos® mix rooms.
This is a near field mix. The most common monitoring levels for near field are 79db or 82db.
Peaks should generally not exceed -2dBFS true-peak. To achieve this, we recommend setting true-peak limiters on all beds and objects at -2.3 or lower. Peaks should be measured via a 5.1 rerender.
The 5.1 M&E rerender is required and must conform to Netflix Peak specification.
Optional tracks are required in addition to the Atmos® M&E when necessary.
All beds and objects are encoded and will always be audible. A printmaster may utilize a composite of all bed material OR multiple beds, but not both.
M&E bed and object configuration should match the Atmos® printmaster. Dialog beds and objects should contain only production effects and no discernable language.
If a 85db reference theatrical mix is created, two complete sets of deliverables are required. One for theatrical, one for near field.
Leader and sync pop are preferred but not required.
M&E Mix
- Dolby Atmos® BWAV ADM File or Pro Tools Recorder Session.
  - All bed channel configurations are acceptable.
  - 48 kHz, 24-bit.
  - Single or multiple M&E Options stems as separate WAV file(s) or marked clearly as Optional or OPT within the Pro Tools session, along with any other content-specific labels.
    - Channel format should be determined by the re-recording mixer to match the intent of the original mix.

2.7.2 Near Field 5.1 Surround M&E Mix

Provide a fully filled 5.1 surround submix containing only Music & Effects (no dialog).
Ambiences and foley fill should be included to correctly match the original mix. Any non-dialog sound present in the main mix must also be represented in the M&E mix.
All levels should mirror what was used for final Mastering and Archive so that any newly added dubbed dialog can be mixed in easily later.
Optional tracks are required in addition to the 5.1 M&E when necessary.
For Dolby Atmos® mixes, the 5.1 M&E should be rendered from the Dolby Atmos® M&E.
M&E Creation Guidelines documents:
- Narrative
- Nonfiction

3 Theatrical Mix - Mastering and Archive

Theatrical Mixes should follow the near field specification, listed above, with the following exceptions:
Use 85db spl as your standard reference level for mixing
Theatrical mixes have no LKFS loudness requirement and may peak at 0db True Peak
Provide Dialog, Music and Effects Stems at Theatrical spec
Delivered as reels (not longplay)

4 Best Practices

NOTE: The below are not Technical Specifications. The below is guidance to help achieve an optimal customer experience. Following the guidelines below is not required by Netflix, but is preferred.

4.1 Dialog-Gated Loudness and Loudness Range

Netflix is committed to protecting creative intent - we do not compress, limit or modify audio mixes. However, content which is not excessively dynamic provides a better experience for members. Please ensure that dialogue is always intelligible within the mix, except where it’s intended to be unclear.

We normalize all audio deliveries to ensure all audio plays at the same level on service. We raise or lower the overall mix level to ensure consistent playback between titles.

The following metering products are tested and known to be effective for ensuring conformance to our audio specifications:

Plug-ins Enterprise Solutions

Dolby Media Meter Dolby DPLM

NuGEN Vis-LM Interra Systems Baton

Waves WLM Emotion Systems Eff

Signum Bute NuGEN AMB

Izotope Insight Minnetonka Audio AudioTools Server

Youlean Loudness Meter

The following loudness range (LRA) values will play best on the service:

5.1 program LRA between 4 and 18 LU
2.0 program LRA between 4 and 18 LU
Dialog LRA of 10 LU or less

When near field Atmos or 5.1 mixes measure at less than 15% dialogue, program target measurement will be used instead (-24db LKFS +/- 2 LU - ITU BS 1770-3 or -4)

For LoRo fold downs or down mixes, ITU-R BS.775-1 - 5.1 to Stereo with LFE muted is the international standard.

4.2 Peaks

We recommend setting a True Peak limiter at -2.3 for all audio deliverables. For 5.1 / 2.0, this helps to prevent false positives from minor differences in metering. For Atmos beds and objects, this setting will reduce excessive peaks due to summing factors within the Atmos renderer.

4.3 Room Setup

We recommend near field mixing in rooms that approximate the size of a living room. Larger or smaller rooms can also work, when they are well-built and aligned. Use multiple microphones near the mix position which average to align rooms. Atmos rooms should be 7.1.4 at a minimum, but 9.1.6 is optimal. We suggest a moderate or light X-Curve for near field, rather than the standard cinema curve. Dolby engineers are available to assist with alignment for facilities that would like it.

Change Log

OC-1-5 (2023-11-29)

Updated to align with changes to Post Production Branded Delivery Specifications.
Separated 5.1 and 2.0 Original Language specifications from Dubbed Languages and Audio Description specifications.
Removed section 2.7.3 Near Field 2.0 Stereo M&E Mix

OC-1-4 (2021-11-09)

In 2.1, removed the requirement that the program must match the final IMF picture, including 1 second of black at head and tail.
In 2.1, added requirement to label all beds or objects as Dialog, Music, Effects or Narration
In 1.0, changed Source Management back to Backlot

OC-1-3 7/1/2021

In 1.0, added line: Deliver all Near Field audio assets as a longplay (not reels).
In 3.0, added line: Delivered as reels (not longplay).

OC-1-2 11/12/2020

In 1.0, changed Backlot to Source Management
In 1.0, added information about leader and Project Files
Clarified that all tracks may be used for objects or beds
In 2.1, added simplified Atmos in IMF delivery information
Added section 2.2 for dubbed and audio description Atmos delivery
Removed 2.0 M&E specification
Added Atmos M&E specification

OC-1-1 4/4/2019

In section 3, removed loudness spec. Following it is required and not optional
In section 3, added current tested loudness meters
In section 3, revised LRA guidance to 18 for full program
In section 3, removed LRA guidance around difference between FX and Dialog
In section 1.2, removed all references to LtRt to explicitly require LoRo
In section 1.5, added preference for LoRo Atmos rerenders
Renamed doc Sound Mix rather than Audio Mix
Added context in Prerequisites that audio deliverables must match camera/picture post frame rate
Added suggestion for a moderate or light x-curve to Best Practices

OC-1-0 7/26/2018