Introduction
This article details Netflix’s expectations for the Dialogue List (DL) and Pivot Language Dialogue List (PLDL).
The Dialogue List and the Pivot Language Dialogue List are used by dubbing studios to create the dubbing script and dubbed audio for their target language. A consistent style for all projects provides a better output for our partners and stakeholders.
All partners creating DL or PLDL in our Authoring tool must follow the below requirements.
Table of Contents
The distinction between DL and PLDL
The distinction between DL and PLDL
A Dialogue List (DL) is the transcription of the dialogue in the original language of the content. When the original language isn’t English, an English translation of the transcribed dialogue is included to serve as a pivot language, hence the name Pivot Language Dialogue List (PLDL).
In essence, a PLDL is a dialogue list with a pivot language.
Disclaimer
For the purpose of this document, the terms Dialogue List and DL are used throughout, as the styles and requirements detailed in the sections below are agnostic of the source language of the content. Any special treatment needed for the pivot language, the English translation, is noted where required.
Elements of the DL
Authoring the Dialogue List consists of transcription, character attribution, and note tagging. When the audio language is not English, a translation into English is added to serve as a pivot language.
Each dialogue event is a contained unit and organized linearly in the dialogue editor section, and each event requires some level of editorial pass.
Event UI in Authoring
|
An event when the source language is not English |
Editable event fields: IN: The in-time of the dialogue. OUT: The out-time of the dialogue. Character: The name of the character speaking the dialogue. Cue: Indicates if the speaker is on-screen or off-screen. Text: Audio or on-screen text transcription. Text (English): Translation, in English, of the dialogue or text. |
|
An event when the source language is English |
Event editorial
Spotting and dialogue
Timecodes
The spotting of the event.
- Timecodes should be frame-accurate to Netflix burn-in timecodes. The timecode is automatically pre-laid for each dialogue event based on automatic speech recognition. For projects where the source language is English, transcription is also automatically pre-laid.
- Adjust the IN and OUT fields as needed.
- We ask that timecodes are within a 3-frame tolerance of the lip flap or spoken word start and end time.
- Adjust the pre-laid transcription as needed to comply with the below requirements.
- Split an event or spot a new one when there is a pause in the dialogue of two or more seconds, even if it’s still the same character speaking.
- When adding new events ensure there’s a two-frame minimum between events.
Characters
This field contains the name of the speaker or character.
- Character names should be Romanized (when their original spelling is not in Latin script) and written in ALL CAPS.
- Spelling of names should be according to credits, on-screen text, or other source materials such as shooting scripts.
- Keep the character name consistent across all episodes.
- For example, don’t use the character’s last name in one episode and in another episode, their first name.
- Keep the character name consistent across all episodes.
- In case of spelling discrepancies among the materials, notify the Netflix representative.
- Characters with generic names (i.e., NURSE, MAN IN SUIT) that have their names revealed or identified in later episodes should be updated to correctly identify them by name in all episodes where the generic name was used. Flag to the Netflix representative all generic characters that have been named.
- Use NONE as the character for on-screen text (e.g., graphics, overlays, principal photography).
- NONE is a reserved character name. As such, it should only be used as indicated above.
- Character names should represent an individual.
- Use one character per event.
- Don’t list characters separated by commas (MIKE, ELEVEN) or slashes (MIKE/ELEVEN) in the same event.
- Do not use character names such as GROUP, VARIOUS, or ALL.
- In case multiple characters say the same dialogue simultaneously, duplicate the event, once for each extra character, and attribute the new events to each character voicing the dialogue.
- For crowd dialogue (such as chants), use the character name WALLA.
- When using a generic character name, as best as possible, identify the perceived gender using the character’s appearance and voice to help dubbing studios identify the types of background voices they’ll need to cast for throwaway lines.
- Always use the word IDENTIFIED in the generic character name and the perceived gender.
- Instead of CHARACTER 1, use IDENTIFIED FEMALE 1.
- Instead of NURSE, use IDENTIFIED MALE NURSE.
- Instead of PA ANNOUNCER, use IDENTIFIED FEMALE PA ANNOUNCER.
Dialogue
The transcription of the original audio or on-screen text relevant to the story.
- The transcription should be accurate and inclusive of all dialogue and stutters or syllables.
- Do not add sound descriptions or reactions.
- E.g.: [sighs], [laughs], [screaming], (long pause).
- Do not transcribe or translate non-diegetic songs unless instructed otherwise by Netflix.
- Non-diegetic songs are songs the audience can hear, but the characters on-screen cannot.
- Utilize dialogue stems and shooting scripts whenever available to clarify/check all prominent dialogue. If these resources are unavailable, reach out to Netflix to confirm if they exist if they would be helpful in the process. Reach out to Netflix if any prominent, indistinct dialogue remains to seek clarity.
Translation
This field is available when the audio language of the content is not English. The translation of the event’s dialogue into English is added to this field to serve as the pivot language.
- The English pivot translation should be a linguistically correct and fluent back translation of the original text.
- The translation should be a natural, comprehensive, and idiomatic rendition of the original text, preserving the tone.
- Do not consider lip sync or other adaptation of the translation.
- Do not change places, brands, film/tv programs, celebrity names or any other cultural reference in the translation. If the term is unlikely to be well known outside the country of origin, please keep the term as is and add the relevant annotation.
Cues
The Cues field indicates if the speaker is on or off screen while speaking.
- The default option is On and indicates the speaker is on-screen.
- When a character speaks off-screen, indicate by selecting Off.
- If the character’s dialogue section alternates between on- and off-screen, indicate the order that it occurs.
- Select Off/On for a character’s dialogue that starts off-screen and continues on-screen.
- Select On/Off for a character’s dialogue that starts on-screen and continues off-screen.
Annotations
The annotations field provides context about the dialogue, character, action, or story. Annotations added to an event in the DL should have one of the pre-defined annotations categories selected.
Annotation
The annotation's purpose is to aid the translator when there is room for misinterpretation and misunderstanding that the context of the content cannot solve. It should bridge any lost context and meaning when the translator has to rely on the pivot language (English).
Use the annotation notes box to provide context.
- Explain word plays, double entendres, and puns lost in the back translation.
- Cultural references that are not explained in the content, and foreigners need to understand the dialogue or action.
- Indicate citations of known quotations (prose, verse, lyrics).
- Slangs.
- Do not add annotations that simply provide a word's definition from the dictionary.
- Do not explain what can be inferred from the dialogue and context of the scene.
- For example,
- The dialogue is “Pangaré here can take us there.”
- Annotation says: “Pangaré is the name of the mule.”
- Context: The speaker is pointing to the mule on-screen as the dialogue is spoken, making it clear that Pangaré is the mule.
- For example,
Annotation categories
The annotation categories function like a tag to classify events and standardize the treatment of different elements in the content, reducing the need for free-form notes in the notes box.
- Use multiple categories only when required. Don’t use more than one if one will do.
- Do not use Context with other categories. If an event is a song lyric, tag it with the appropriate category and leave any annotation in the notes box.
- Always select a category when leaving an annotation in the note box.
- Select the appropriate category according to their intended usage, as described below.
- All events in the DL eligible for an annotation category must be appropriately tagged.
Annotation category types
Each annotation category type has an intended usage, described below.
Context
- Use it to clarify the dialogue’s meaning and context that translators need to do a proper rendition of the dialogue in their target language.
- Always include the annotations in the notes box (not the dialogue box).
- See the Annotation section above for a list of what needs more context.
On-screen text
- On-screen text that is prominent and relevant to the story should be spotted in the DL and tagged.
- Include the transcription of the on-screen text in the dialogue event box (not in the annotation box).
- For languages that capitalize letters, the transcription of the on-screen text and its English translation should be in ALL CAPS, except for long passages of on-screen text (e.g., prologue or epilogue), which should use sentence case to improve readability.
- The event spotting should match the duration of the on-screen text while observing the minimum and maximum duration allowed.
- Minimum duration: 20 frames
- Maximum duration: 7 seconds
- When on-screen text fades in or out, experiment with different timings and aim for sync between the event and the on-screen text.
- Limit lines to a maximum of 42 characters when transcribing English text or translating the on-screen text into English (the pivot language).
- Do not use more than two lines of text in the dialogue box.
- For all source languages and the English translation (when applicable).
- If the text won’t fit in two lines, add another event while still respecting the minimum and maximum duration allowed and matching the duration of the text on-screen.
- Indicate the on-screen region where the event containing the on-screen text should go.
- By default, the event’s text will display at the bottom of the screen.
- Select the top region when the event obscures text in the lower third of the screen.
- Do not toggle the Forced Narrative switch to on.
- Use the appropriate sub-category for events tagged as on-screen text.
- On-screen text sub-categories:
- On-Screen Text - Burned-in subs
- Use it for transcriptions of production-approved burned-in subtitles as they appear in the current cut of the picture.
- Content/timing of events in the DL should match the most recent cut or document provided by production. Line and subtitle breaks should be honored as they appear in documents.
- Please split continuous dialogue into rows according to corresponding subtitles if applicable. If the timing of dialogue audio and burn-in subs are not the same, spot the events according to dialogue.
- If burned-in subtitles contain content translating foreign dialogue, indicate the language spoken in the notes box (do not include it in the dialogue box).
- For hard-to-hear dialogue with burned-in subtitles, the dialogue event should still contain a transcription that matches the content in the burned-in subtitle.
- Note: Final picture will often be clean of burn-in subtitles, but this content should stay in the final Dialogue List and be timed to match the final version of the picture as it would occur had it been left in.
- On-Screen Text - Graphics & Inserts
- Use it for graphics added during post-production, such as location cards, text messages, dedications, prologues, and others.
- On-Screen Text - Principal Photography
- Use it for on-screen texts in the same environment as the characters.
- On-Screen Text - Episodic Title
- Use this subcategory when the episode title appears on-screen.
- If the episode title is voiced-over in the cut, treat it as dialogue (not as on-screen text), but still tag it as On-Screen Text - Episodic Title.
- Do not translate the episode title freely when adding an English translation to serve as the pivot language. Always use the approved English translation found in Terminology. If there isn’t an approved English translation, enter EPISODE TITLE in the translation box.
- On-Screen Text - Netflix Credits [DO NOT USE]
- Do not spot any event for Netflix Credits unless instructed otherwise.
- On-Screen Text - Main Title
- Use this subcategory when the main title of the content appears on-screen.
- Do not translate the main title freely into English when adding an English translation to act as the pivot language. Always use the approved English translation of the main title found in Terminology. If there isn’t an approved English translation, enter MAIN TITLE in the translation box.
- On-Screen Text - Burned-in subs
Foreign dialogue
- Transcribe and italicize words and short phrases in the dialogue box that are different from the original version’s language.
- Indicate the spoken language in the notes box (not in the dialogue box).
- When translating the dialogue into English for the pivot language, keep the foreign words and short phrases spoken transcribed in their language and italicized.
- Any foreign dialogue creatively intended to be understood (i.e., subtitled in the original version) should be translated into English.
- When the source language is not English, any dialogue spoken in English should be treated as foreign dialogue.
- For longer passages that cannot be transcribed without the assistance of a native speaker, flag them to Netflix.
- If any mistranslations are noted in burned-in subtitles, flag them to Netflix before updating the DL.
- The event spotting should match the audio while observing the minimum and maximum duration allowed.
- Minimum duration: 20 frames
- Maximum duration: 7 seconds
- Limit lines to a maximum of 42 characters when transcribing the audio or translating it into English (the pivot language).
- Do not use more than two lines in each event.
- For all source languages and the English translation (when applicable).
- If the text won’t fit in two lines, add another event while still respecting the minimum and maximum duration allowed.
Songs & Ditties
- Use this annotation category on events that contain transcriptions of vocals, whether licensed, unlicensed, original IP, or ditties.
- Soundtrack, including opening and closing theme songs, should not be transcribed or translated unless instructed otherwise by Netflix (see also the Dialogue section).
- For Kids & Family titles, please include transcription of the opening and closing theme songs by default.
- Soundtrack, including opening and closing theme songs, should not be transcribed or translated unless instructed otherwise by Netflix (see also the Dialogue section).
- Italicize the song lyrics in the dialogue box and the translation box if adding an English translation for pivoting.
- The event spotting should match the audio while observing the minimum and maximum duration allowed.
- Minimum duration: 20 frames
- Maximum duration: 7 seconds
- Limit lines to a maximum of 42 characters when transcribing English lyrics or translating them into English (the pivot language).
- Do not use more than two lines in each event.
- For all source languages and the English translation (when applicable).
- If the text won’t fit in two lines, add another event while still respecting the minimum and maximum duration allowed.
- Do not mix spoken dialogue and singing in the same event.
- Use only question and exclamation marks at the end of a line for song lyrics (no commas or periods), both in the transcription and the English translation when applicable.
Archival Footage
- Use this annotation category for anything filmed outside of the film/series, such as TV news footage, court footage, previously released movie/TV content.
- Transcribe the dialogue and translate it into English if adding a pivot language.
- The event spotting should match the audio while observing the minimum and maximum duration allowed.
- Minimum duration: 20 frames
- Maximum duration: 7 seconds
- Limit lines to a maximum of 42 characters when transcribing the audio or translating it into English (the pivot language).
- Do not use more than two lines in each event.
- For all source languages and the English translation (when applicable).
- If the text won’t fit in two lines, add another event while still respecting the minimum and maximum duration allowed.
Existing IP
- IP (Intellectual Property) refers to creations of the mind. It refers to everything from works of art to inventions, computer programs, trademarks, and other commercial signs.
- Identify any instance of IP as accurately as possible. If unable to identify the IP, reach out to Netflix.
Inaudible
- Use this category to indicate Instances of inaudible dialogue.
- In cases where there are any discernible words across a mostly indiscernible passage, please include them as opposed to leaving the whole event as inaudible.
- Ensure the dialogue stems are unavailable before tagging an event as Inaudible (see Dialogue section above).
- Flag to the Netflix Dubbing representative projects that contain inaudible dialogue.
Main Title in Dialogue
- Use this tag for any instance of Main Title usage in dialogue.
- When creating the English translation for the pivot language, always refer to the approved translation in Terminology.
- If the approved English main title translation can’t be worked out in the dialogue translation, translate it as faithfully to the approved version as possible to obtain a coherent sentence, trying to preserve the intended meaning of the version found in Terminology. Leave a comment in the notes box indicating that the translation doesn’t match the approved version and flag it to Netflix.
Episode Title in Dialogue
- Use this tag for any instance of episode title usage in dialogue.
- When creating the English translation for the pivot language, always refer to the approved translation in Terminology.
- If the approved English episode title translation can’t be worked out in the dialogue translation, translate it as faithfully to the approved version as possible to obtain a coherent sentence, trying to preserve the intended meaning of the version found in Terminology. Leave a comment in the notes box indicating that the translation doesn’t match the approved version and flag it to Netflix.
- If there isn’t an approved translation for the episode title, i.e., it matches the original version, translate the episode title into English as closely as possible to the intent of the original version. It should still be tagged, and leave a note in the notes box indicating that it’s a free translation.
Sensitive Context
- Use this annotation tag category on events containing words or phrases related to gender or gender identity, race, ethnicity, LGBTQIA+, disabilities, religion, cultural sensitivities, profanity, or other categories that could be perceived as offensive or insensitive in the given context.
- Refer to the Sensitive Terms library in Terminology for a list of sensitive words and their severity level.
- The list of words in Terminology is extensive but not exhaustive.
- Flag words that can be viewed as sensitive even if not found in Terminology.
- Provide relevant information about the word and subject in the notes section.
- Add context in the notes section around the severity of the expletive within its context, as this may not be clear from the English translation.
- When adding a pivot translation in English, use an approved equivalent found in the Terminology tool. If the word isn’t listed there, use a translation with the same meaning and severity as the original version.
Character editorial
- The characters list should be free of unused characters due to importing the shooting script.
- Remove any characters that have no dialogue.
- Each character in the list should be one and only one individual.
KNP entries creation
Transcribers and translators are responsible for creating KNP (Key Names and Phrases) terms in Terminology.
- Add key source terms and phrases in the original language following the KNP Source Term Creation & Translation Guidelines
- English translations for all original language terms when the source language isn’t English.
- The KNP is accessible from the Authoring application via the reference materials tab
Image 1: Reference Material tab
- KNP source term creation is due within the DL timeline.
Reference materials
- Dialogue List Scope of Work
- Pivot Language Dialogue List Scope of Work
- Terminology Tool Overview: Key Names and Phrases (KNP)
- Script Authoring Tool User Guide