Delivery & QC

A Delivery Kick Off meeting is to be scheduled prior to the delivery of any cuts. The purpose of the meeting is to review title tech specs, asset specs, file naming conventions, upload destinations, QC procedures, and related workflows.

Delivery Contacts

Available here if you are logged in.

Delivery Checklist/Schedule

Delivery Checklist/Schedule will be provided by AMS contact and it contains the complete list of deliverables a production is required to turn over. The checklist/schedule will be determined ahead of budget lock to ensure all deliverables are accounted for in the budget. Production is to use this as a guide to confirm that all deliverables have been submitted for delivery. Once all deliverables are received, a title can be considered wrapped. Contact the AMS Post/Delivery Executive if you have any additional questions.

Production Masters

Production Masters are the sources used to create streaming files for Prime Video distribution. AMS requires delivery of Production Masters in both HDR and SDR dynamic ranges. If a title is delivering in HD, then only SDR is required (Animation).

There are three different types of Production Masters:

Texted: Contains main titles, end credits, and any other instance of text. Forced Narrative subtitles may or may not be present for events where characters are speaking a non-OV language.

Non-Subtitled: Do not include burned-in Forced Narrative subtitles. Utilized for platform distribution which will ultimately be seen by customers.

Textless: Required when 30% or more of the program is texted; in these cases, a Texted Production Master must also be delivered.

Textless Elements must be frame accurate to Production Master. Utilized for platform distribution which will ultimately be seen by customers.

An HDR Report must be provided to the QC vendor at the time of HDR delivery to ensure the color and luminance values are to spec. HDR Reports can be generated from a Colorfront system from your finishing vendor.

Textless Element Reels

A Textless Element Reel is required for each movie/episode and must be delivered as a separate file. A textless element needs to be provided for every instance of text over picture, graphics elements with text, and animated text. Examples include but are not limited to:

• Animated Text
• Cards
  • Disclaimers
• Chyrons
• Credits
• End Titles
• Forced Narrative Subtitles
• Graphic Elements with Text
  • Cell phone messages
  • Infographics
• Locations
• Main Titles
• Intertitles
  • A word or group of words that appear on-screen during a movie/show but are not part of a scene.
• Time/Date
• Visual Effects

AMS recommends that textless elements are built from a final online sequence to ensure accuracy. Textless elements are not required for backgrounds that are solid black or white.

A Textless Production Master is required when 30% or more of the program is texted, in these cases a Textless Element Reel is not required.

Dolby Vision

For titles finished in Dolby Vision, Amazon Studios recommends that each production reference the Best Practices Guide: Dolby Vision Color Grading. When starting a project select the best available files to preserve the dynamic range and color information. Dolby Vision deliverables should be created using Full Range Video/Data levels. Please confirm with your Post Executive if Dolby Vision is required.

1. Camera RAW - Best
2. Uncompressed 16-bit, RGB, 444 - Good
3. Uncompressed 12-bit, RGB, 444 - Acceptable

Legal Range deliverables, like a UHD SDR Production Master are to derive from the Full Range Master, taking care to scale the image correctly during the conversion and avoid any clipping or loss of detail in the highlights and shadows. 

Dolby Vision Metadata is a sidecar XML file that is frame-synced to the TIFF image sequence that is used to create Dolby Vision Master.

If a title contains Forced Narrative subtitles, then a UHD Dolby Vision Production Master - Non-Subtitled must be created. Forced Narratives are not to be included in any Dolby Vision Production Masters.

Split Track Masters

Split Track Masters are utilized for marketing purposes. Master will include the program with textless element reel at tail, Printmaster audio, and Dialogue/Music/Effects audio stems. Refer to section Post – Asset Technical Specification - Split Track Master for asset specifications.

Near Field Loudness Levels

Follow the Loudness guidelines noted below for Near Field 5.1/2.0 mixes only. Does not apply to Dolby Atmos audio. Refer to the Post – Asset Technical Specifications - Audio section for asset specifications.

Amazon Studios recommends that the 1770-1 (with Dialogue Gating exclusion algorithm/Dolby Dialogue Intelligence) loudness scales is measured throughout the mixing process.

The Loudness specifications provide parameters to ensure consistent audio levels and optimize customer experience when streaming content.

Amazon Studios reference measurement model is Nugen Audio VisLM2. VisLM2 Amazon Studios preset files are available here.

QC Guidelines for Loudness

QC Facilities should measure loudness with the following scale:

• BS 1770-1 loudness scale with dialogue gating exclusion algorithm/Dolby Dialogue Intelligence 

Include the values for the loudness scale for 5.1 and Stereo within the QC Report.

• If the 1770-1 loudness scale with dialogue gating exclusion algorithm/Dolby Dialogue Intelligence is not within -27 +/- 2 LU range, please flag as a rejection on the QC report.

Dolby Atmos

When creating a Dolby Atmos mix for a title AMS prefers that all channel-based audio to be derived from the Dolby Atmos Printmaster or Recorder Session. A Near Field Dolby Atmos Printmaster ADM WBF file is required for Prime Video Distribution and must be put through a QC and receive a passing QC report. Dolby Atmos audio does not have a loudness standard requirement.

M&E

All M&E tracks should be fully-filled and include all music (Source, Score, etc.), sound effects (SFX), room tone (Ambience/Backgrounds) and Foley as they occur in the domestic mix. Any production effects (PFX) that overlap with dialogue must be replaced/covered with cut SFX or Foley.
Note: Non-fully-filled M&E files will automatically be rejected for fixes.

Please see additional Unscripted, Documentary, and Animated Program requirements below.

Elements to Include in the M&E

• Kissing, grunting, groaning, coughing, sneezing, breathing, reactions, or other noises / vocal intonations not discernible as specific language made by principal or ancillary characters who do not speak / have no adjacent dialogue lines.
  • Note: These elements above should only be placed in the M&E if they are sourced from an SFX library or are “in the clear” and do not have any vocal tone attached to them.
• Generic crowd noises, walla, groups, cheers, or reactions with no discernible language or dialogue of any kind.
• Background/ambience and room tone beds.
• Baby crying if the baby has no discernible dialogue or is from an SFX library.
• Sourced (licensed footage) clips with audio (e.g., movies, TV shows, radio broadcasts, commercials) without discernible language or dialogue (see below if dialogue is present).
• Instrumental music separated from any performance/cast vocals. Live “on set” instrumental performances must be re-recorded clean of vocals.
  • Note: If performances are recorded live, and vocals are married to the accompanying music, the instrumentals need to be re-scored/re-recorded free of vocals and included in the M&E. These instrumentals must match the domestic mix(es) levels, treatments, fades, and tone. The original married performance should also be supplied “as-is” in an Optional stem (See Optional Stems).

Elements Not to Be Included in the M&E

• Discernible dialogue of any kind. This includes foreign language and discernible dialogue in groups or walla. Unless approved by Localization due to creative reasons, discernible dialogue lines should always be removed, replaced, and/or obscured with generic “clean versions” in the M&E. Please place the elements with distinct dialogue in the optional tracks.
• Vocals or non-verbal sounds made by main or ancillary characters adjacent to, or overlapping, dialogue lines (e.g., kissing, grunting, groaning, coughing, sneezing, breathing). These elements will likely be re-recorded by foreign language voice talent (See Optional Stems).
• Dialogue from licensed media (sourced for movies, TV shows, radio broadcasts, commercials, etc.) when possible.
• Dialogue that has been scripted/created for playback as a movie, TV show, radio broadcast, commercial, etc.
• Discernible futzed dialogue (Intercoms, PA announcements, radio “chatter”, TV, phones).
• Reverb sounds or dialogue “treatments”.

Optional Stems

Optional stems should be delivered along with the M&Es and can be provided across multiple sets of stems when the events overlap or occur simultaneously.

AMS requires the delivery of M&E/optional track mapping documentation if available.

Materials should be provided at the same time the event occurs during program, meaning the Optional stem(s) should run the entire duration of the feature.

All program material in the Optionals should play in sample accurate sync with Printmaster and not create any phasing when played with the M&E.

Any program material in the Optionals should include the treatment heard in Printmaster (i.e. futz, reverbs, panning).

Elements to Include in the Optionals (If Applicable to Title):

• All on-screen performances or cast vocals (clean/separated from accompanying music).
• If performances are recorded live, and vocals are married to the accompanying music, the instrumentals must be re-scored/re-recorded free of vocals and supplied in an Optional stem. These instrumentals must match the domestic mix/es level, treatments, fades, and tone. The original married performance should also be supplied as-is in an additional Optional stem.
• If two or more characters are singing at the same time, vocals should be split-out into two optional stems (if recorded separately).
• Non-OV (Foreign) language lines (even if the non-OV language is spoken by a character who speaks the OV language). Overlapping foreign dialogue lines by different characters need to be placed in separate optional tracks, if possible.
• Walla or group recordings with discernible foreign dialogue lines.
  • Note: This should only be done when the foreign dialogue in the walla is “story relevant”. The default approach should be to provide “clean” walla in the M&E whenever possible.
• Kissing, grunting, groaning, breathing, sneezing, coughing, reactions, or vocal intonations from speaking characters that have been omitted from the M&E, provided they do not overlap dialogue.
• Source (licensed footage clips) with audio that includes discernible language should be provided AS-IS in one set of Optional stems.
• An additional Optional stem should also be supplied to replicate the sourced footage, free of any discernible dialogue. Any MX, FX, video ambiance, etc. that was married to dialogue in the source footage (i.e., audience clapping or laughing, doorbells, TV static, commercial jingle music, etc.), should be recreated with suitable sound-alikes.

Final Stems

Stems are to be provided for each audio element (dialogue, music, effects, etc.) in the applicable audio configurations. Refer to section Post – Asset Technical Specifications - Audio for more details. Stems are to be delivered to the Localization vendor as part of the Localization Audio package (Nearfield Printmaster, M&E, and Stems sessions). For archival QC, stems are to be delivered to AMS after Production Master and M&E have both passed QC.

As-Broadcast Scripts (ABS), Closed Captioning (CC), Audio Description (AD), and Quality Control (QC) are the responsibility of the production. ABS are required by the time that production masters are provided to the QC vendor. Quality Control must include in-scope all Original Version Masters, CC and AD files and must not be conducted by the same vendor(s) that creates CC or AD files. Quality Control of the CC files must be a watch down of the entire file for caption accuracy. AD files should be a 5-point check for sync. If timing does not allow for AD and CC quality control by a different vendor (e.g., for live or near-live shows), reach out to the Post Executive for approval to bypass this.

Find Timed Text Guidelines here.

Find Language Specific Time Text Guidelines here.

Forced Narrative Subtitles

Forced Narrative (FN) subtitles translate spoken dialogue and on-screen text that is in a different language from the primary audio content of the program when creative intent requires that the dialogue be understood by the viewer. They are displayed to the customer based on their audio choice, rather than selecting that subtitle language. If a title contains Forced Narrative content, delivery of separate Forced Narrative subtitle assets is required. 

SDH Subtitles

Subtitles for the Deaf and Hard-of-Hearing.

Full SDH Subtitle: Includes Forced Narratives.

Sans Forced Narrative SDH Subtitle: SDH subtitle without Forced Narratives.

Link to: Amazon SDH and CC Style Guide_032024_V1.4.docx 

Audio Description

Audio Description (AD) is the verbal narration of key visual elements in television, movies, and other media to enhance the blind or low-vision consumer’s experience. It is also known as VI (Visually Impaired track), NAD (Narrative Audio Description), and DVS (Descriptive Video Service). AD is inserted in between plot-pertinent dialogue, sound effects and/or music to describe settings, characters and actions that will enable a better understanding of the story. While AD is primarily used by persons who are blind and/or visually impaired, it is also known to benefit a wide range of customers, including auditory learners and neurodiverse audiences. Vendors must be provided with screener, Production Script, and Printmaster audio to create AD.Amazon requires the following AD deliverables to be created:

• Audio Description Script
• Near Field Audio Description 2.0 or 2.0 + 5.1
• Near Field Audio Description Session

Link to: AD_Style_Guide_Amazon_Studios_122023_V1.4.docx

A Digital Cinema Package (DCP) is utilized for Theatrical distribution and exhibition. A DCP is a collection of audio, image, and data files. AMS Post Executive will advise when a DCP is required and provide the format specifications.

Distribution Key Delivery Message (DKDM)

A DKDM is a security passcode that allows DCP's to be decrypted within a mastering system. A DKDM must be created for each DCP.

Key Digital Management (KDM)

A KDM unlocks screening encryption so that movie can be shown in theaters. Targets DCP unlocking during a specific window of time for exhibition purposes. If a KDM is not provided the theater will not be able to be shown.

Digital Cinema Package (DCP) Reel Screener

DCP Reel Screener serves as a reference of the Digital Cinema Package (DCP). Any Forced Narratives or subtitles must be burned into picture. Used by accessibility vendor to create theatrical subtitles and audio description.

A Master QC is a 100% QC of all video, audio, and on screen text of the Master file. It ensures the file is technically accurate, follows AMS specs and highlights any production issues needing review and possible fixes.

Technical QC

A Technical QC validates the file is good and the technical specs are accurate throughout, including all video and audio. It does not note production issues.

Spot Check QC

A Spot QC is the minimum-level QC and does not typically involve the generation of a QC report because the scope of technical validations is so narrow. The spot QC is used on assets derived from sources that have already undergone a higher level of QC evaluation.

The Spot QC is intended to surface any significant issues induced through the media transformation process, including incorrectly mapped audio tracks, media corruptions, assembly errors, audio/video sync, video compression or artifacting issues, etc.

Spot checks can be either a Three Point Check or a Five Point Check.

• Three Point Check - 0-30 minute TRT (Total Run Time)
  • First 2 minutes
  • 1 minute at 50% program
  • Last 2 minutes
• Five Point Check - 31+ minutes TRT
  • First 2 minutes
  • 1 minute at Main Title sequence
  • 1 minute at 50% program
  • 1 minute at 75% program
  • Last 2 minutes

Titles may need to pass a qualitative analysis known as the “Harding Test” or “Harding Analysis” to check the program for instances of photo sensitivity that could cause seizures for viewers.

Kids, Animated Series, and All UK Titles

A Harding Test is required for all kids and animated series, and UK titles. For any other content, please consult your AMS Post Exec to determine if a Harding Test is required.

If the title fails the Harding Test then the Post Exec will attempt remediation and run another test. If the content continues to fail after attempts to fix it have been made, then the content will be sent to the platform with a disclaimer card.