French Guidelines For Transcriptions and Subtitles [PDF]

Zitiervorschau

French Guidelines for Transcriptions and Subtitles – frBE, frCA, frCH, frFR 

Scope



1. General Information



2. Flavour



3. Style & Tone of Voice



4. Text Reduction



5. Interjections



6. Consistency



7. Special Content



8. Special Formatting



9. Language-specific Items – French



10. Terminology



11. Notation Tags



12. Transcription Specifics



13. Subtitles Specifics



14. Useful Tools



15. Reminders



16. Version Log

💡 🏄🏿‍♂️QUICK-SURFING HACK🏄🏼‍♀️The CTRL+F (or Command+F) function works in open toggle lists. If you want to quickly close all the toggle lists to check the list of the sections at a glance click on CTRL+Alt+T (Command+Alt+T). Each section is completed with a list of search keywords in your language to facilitate your navigation. ‣

Scope

Welcome to Happy Scribe Guidelines, a set of rules and hacks for granting consistency and delivering high-quality transcription and subtitling projects.   ✨Take the most out of them! ✨ ‣ How to read the Guidelines? Sections 1-8 → Generic information applicable across languages Section 9 → Language-specific rules Section 12 → Transcription specifics Section 13 → Subtitles specifics ⚙ Looking for extra information or some useful tips? 💡 introduces further details on a topic. 💯 introduces useful hacks. 🔑 introduces localized search keywords. ‣ How to use the Guidelines? 1st time → Check all the sections thoroughly before starting working on your first assignment. Following times → Use CTRL+F (or Command+F) to find the solution to your issue. 💡 You are not expected to read the Guidelines every single time you work on a file. However, you are expected to be aware of the latest additions all the time. The latest additions will be highlighted in pink from time to time. 💯 Quality tip: Use the guidelines efficiently. Are you working on a file with no audio? Run a search for “no audio” and you will know what to do with that. ‣ How to improve the Guidelines? The Guidelines have been designed to be a dynamic tool to fulfil your needs. Feel free to share your feedback to expand, enrich and improve them. Add a reliable source reference any time you make any suggestions. 💡 Share your feedback in the Discord Channel of your language. Please note that when customer-specific instructions are available, they prevail over these Guidelines and any other rules. ‣

   

What languages do these Guidelines refer to? 🇫🇷 🇨🇦 🇧🇪🇨🇭 The following Guidelines include specifications for the following flavours: French for France French for Canada French for Belgium French for Switzerland Flavour-specific rules will be noticed accordingly. If no flavour is specified, the rule is meant to be applied across all variants. ‣

1. General Information ‣

i. Service Level

What type of transcription/subtitles does Happy Scribe offer at the moment? Happy Scribe does not provide Verbatim transcription at the moment. We currently offer clean-read transcripts and subtitles. A clean read transcript aims to make the text legible, concise, and clear, without changing the meaning or definition of speech, or the speaker's intention. Repetitions, stutters, or false starts are left out. In the case of repetitions, they are allowed if the speaker is intentionally repeating certain words for emphasis. The speech continuity should be granted by using the proper punctuation sign. 💯 Quality tip: Check the “Text Reduction” section for further details on how to edit the text accordingly and “Punctuation” to know how to report long pauses. 💡 If the file you are transcribing or subtitling has some embedded subtitles in your language, please DO transcribe the speaker’s speech. DON’T mirror the embedded subtitles even when they are slightly different from the speech spoken by the speaker. 🔑 Search keywords: verbatim, type of transcription ‣

ii. No Speech in File or Poor Audio

You claimed a file but it:  Has no audio.  Has very poor audio. What should you do? 1. Post the file name on the channel unworkable-projects on Discord. 2. Report the file as “unworkable” on the Editor and select either “No speech: there is nothing to transcribe” or “Inaudible: it is impossible to transcribe because the audio cannot be heard”.

3. Unclaim the job. Never submit it as “Done”. 💡 If a file with no/little/very poor audio is submitted as “Done”, no payment will be processed. 💡 If there is very little audio in the file (less than 20% of the total length), please flag it to the Happy Scribe team.

🔑 Search keywords: no speech, poor audio, unworkable, bad audio, mauvaise qualité audio, pas d’audio, infaisable ‣

iii. Songs

You claimed a file but more than 80% of the audio is a song or lyrics only. What should you do? 1. DON’T transcribe the lyrics of the song and DON’T submit the file. 2. Post the file name on the channel unworkable-projects on Discord. 3. Report the file as “unworkable” on the Editor and select “Song: the majority of the audio is a song (80%+)”.

4. Unclaim the job. 💡 If less than 80% of the audio is a song: - If the song takes less than 10 seconds, transcribe the words to the song. Use italics. - If the song takes more than 10, use the proper tag. Check the section “Notation Tag” below to know how to use the right one! 🔑 Search keywords: song, singing, lyrics, unworkable, chanson, paroles, infaisable ‣

iv. Foreign Languages Scenario

DO

The file contains at least 80% of foreign language in the spoken part.

Report the file by selecting “Foreign language[...]”. Unclaim the job.

The file contains both French and a Please don’t copy the subtitles. Transcribe the foreign language but the video has French French part only. subtitles embedded for the foreign part. The file contains both French and a foreign language but without embedded subtitles.

Don’t transcribe the foreign part. Use the proper tag and transcribe the French part only in both transcriptions and subtitles.

💯 Quality tip: If the speaker says “Les enfants jouent avec le... Comment on dit ? Le qìqiú.” when they mean “Les enfants jouent avec le ballon”, DO type it as “Les enfants jouent avec le... Comment on dit ? Le [langue étrangère 00:00:00]”. DON’T type “Les enfants jouent avec le ballon”. In this example, the interviewee is using a word from his native language that is not known to the interviewer: it is not a case of code-mixing or loan word, but a case of communication loss. Even if you were to

speak Chinese and know the word, please don’t add it as it is not relevant in this specific context. 🔑 Search keywords: foreign language, foreign speakers, foreign tag, langue étrangère, mots empruntés, emprunt langue étrangère ‣

2. Flavour

Depending on the geographical area in which they are spoken, some languages can have locales, also known as “flavours”. In Dutch language, for example, there are two flavours: Dutch for Belgium and Dutch for the Netherlands. Always follow the flavour as selected by the customer and displayed on your job dashboard.

💡 DON’T confuse the flavour with the accent. The speaker can be speaking Dutch for Belgium with an English accent.

💯 Quality tip: Before claiming your job, check the flavour requested. This will help you anticipate some of the challenges of the file you will be working on and apply the relevant linguistic conventions for that specific flavour, like the appropriate spelling and punctuation preferences, accent, and other factors. 💡 Loan words and flavours - frBE 🇧🇪, frCH 🇨🇭 and frCA 🇨🇦 Speakers of the above flavours of French are from countries where French stands alongside other languages, leading to a lot of everyday code-mixing with English (frCA), Dutch/Flemish and German (frBE) and German, Swiss German and Italian (frCH). If customers are given the possibility to select these when uploading a file, then they are entitled to expect to get a transcript that would reflect accurately the flavour. When you detect some code-mixing that belongs to the flavour, try and search the Internet for the right word. Don’t forget also that you are part of an international team at Happy Scribe. Use and abuse that amazing chance: ask and tag fellow foreign Scribes on the 🙋-help-and-tips channel on Discord. If after all this, you still can’t find the word, DON’T make wild guesses and go for a [langue étrangère 00:00:00] tag. 🌐 For help (specific vocabulary + loan words), you can check the following (nonexhaustive) glossaries: 🇧🇪 frBE → https://fr.wikipedia.org/wiki/Français_de_Belgique 🇨🇭 frCH → http://mapage.noos.fr/r.ferreol/langage/fr-ch.htm 🇨🇦 frCA → https://www.je-parlequebecois.com/lexique.html et https://fr.wikipedia.or/wiki/Lexique_du_français_québécois 🔑 Search keywords: locale, flavour, localisation, adaptation, accent, specificités locales, variations regionales, romandismes, belgicismes, québécismes ‣

3. Style & Tone of Voice

The style and tone of voice of speakers should be reported faithfully. This doesn’t refer to the intonation, but rather to the usage of a formal or informal language (i.e., the register). Stick to the following pillars when working on a file: Be concise

Avoid long sentences/paragraphs. Look for natural points in the speech, like shifts in topic, emphasis or references and start new ones.

Be Never add content, paraphrase, or rearrange the order of speech. accurate

Be proper

💯

Oral French has a lot of elisions, omissions and contracted forms. They are marks of orality rather than marks of style and tone. In a clean transcript, please restore proper grammatical forms for the following : - contracted forms such as “t’as”(→ ”tu as”), “j’vais” (→ ”je vais”)... - often omitted “ne” in negative sentences (”il va pas venir” → “il ne va pas venir”). - often omitted “il” in “y’a” (→ “il y a”) or “faut” (→ “il faut”) For subtitles, however, contracted forms can be used for the sake of reducing CPS, if it fits the tone of the file.

Quality tip: Before completing your job, run a quick search looking for unnecessary filler words using CRTL+F (or Command+F). 🔑 Search keywords: contractions, contracted forms, informal tone, formal tone, formality, ton, registre, langue orale, marques de l’oralité ‣

4. Text Reduction 

Unless otherwise specified, DON’T use unnecessary filler words, feedback words, interruptions, false starts, and repetitions. DO

DON’T

Stutters & Stammers

Persifleur ! Tu as sifflé ton dernier sifflement.

Persifleur ! Tu as s-s-sifflé ton dernier s-s-sifflement.

Unnecessary filler words

La chaussure était trop petite pour elles, et les sœurs de Cendrillon, elles étaient super en colère.

La chaussure était trop petite, euh, pour elles, et les sœurs de Cendrillon, ben elles étaient super en colère, quoi.

Feedback words (words as ”oui”, “d’accord”... that are more a proof of attention than an actual answer.)

-Elsa n’enlève jamais ses gants. Je pensais juste qu’elle n’aimait pas la poussière.

-Elsa n’enlève jamais ses gants. -D’accord. -Je pensais juste qu’elle n’aimait pas la poussière.

Self interruptions followed with correction

Si l’océan m’a choisie pour retrouver le cœur de Tefiti, c’est qu’il pense que je suis capable d’y arriver.

Si l’océan m’a choisie pour retrouver le cœur de Tefiti, c’est qu’il pense que je peux... que je suis capable d’y arriver.

False starts

J’ai toujours rêvé d’aller voir Je rêve... J’ai toujours rêvé les lanternes pour mes 18 d’aller voir les lanternes pour ans. mes 18 ans.

Repetitions (for fluidity, don’t Aurora avait soudain très Aurora avait soudain très, très keep if casually repeated. Keep envie de poser le bout de son envie de poser le bout de son ONE if insisted on) doigt sur ce fuseau. doigt sur ce fuseau. 💡 What are filler words? Filler words are all those little words and sounds that you add orally but aren’t carrying any actual meaning (”euh”, “ben”, “voilà”, “quoi.”, ”tu vois”, “en fait”, “du coup”, “et donc”...) When a word is used as oral punctuation (or verbal tic), adds no meaning and hinders fluidity of the speech, it can be removed. frCA 🇨🇦 has its own filler words that you need to identify as such and equally remove (”comme”, “tsé”, “faque”, “là.”, “dans le fond”...) 💡

What if an “unnecessary word” is the only answer of a speaker to a question? In that case, the word cannot really be considered as unnecessary and it should be included in the text. DO

DON’T

Russel: On peut le garder, s’il vous plait ? Carl: Non, il va nous encombrer. Russel: Mais c’est un chien qui parle. Carl: Mmmmmm. Russel: Merci monsieur Fredricksen. Carl: J’ai dit non.

Russel: On peut le garder, s’il vous plait ? Carl: Mmm-mmm. Non, il va nous encombrer. Russel: Mais c’est un chien qui parle. Carl: Mmm-mmm. Russel: Merci monsieur Fredricksen. Carl: J’ai dit non.

Text in red should not be included, text in green should be. These are the accepted spellings for the following: Affirmative answers: Mmm-hmm Negative answers: Mmm-mmm 💡 If conjunctions don’t really work as conjunctions but rather as fillers, please remove them. The same applies to any other connectors (relatives pronouns, adverbs, etc.) when they don’t really work for the purposes they should. DO

DON’T

-La belle-mère de Blanche-Neige est jalouse -J’ai regardé La Belle et la Bête avec Emma de sa beauté. -Donc elle a ordonné au Watson hier. -Donc, tu en as pensé quoi ? -C’est Chasseur de la tuer ? un chef-d’œuvre. The conjunction here doesn’t really connect the The conjunction connects the two sentences sentences grammatically speaking. It should be in this case. omitted. 

Truncate further content exclusively where reading speed and synchronicity to the audio are causing you issues in subtitles. Never omit meaningful content. 💯 Quality tip: If you spot an unnecessary filler word, run a quick search throughout the file by using CTRL+F (Command+F) functionality and remove them all. Interjections should not be treated as filler words. Check out the relevant section below for further details. 🔑 Search keywords: text reduction, filler words, stutters, stammers, repetitions, interruptions, nettoyage, mots de remplissage, faux départs ‣

5. Interjections

As part of speech, interjections are exclamations that express speakers’ reactions or feelings. They are grammatically independent of the rest of the sentence where they appear. Only the interjections below are allowed in transcriptions and subtitles: Ah bon ? Hélas !

Ouh là là ! Ouf ! DON’T use any interjections currently not included in the list or the reference. DO reach out to add new interjections that you consider missing. In the case of subtitles, interjections can be omitted to meet character constraints, if their omission does not imply any loss in meaning. 💡 Exclamation marks can be used only for interjections and shouts. Check the relevant section about punctuation to discover when they are forbidden. 💯 Quality tip: How to distinguish filler words from interjections? If the phrases work as a reaction from the speaker, they should rather be considered interjections. Unless you have been instructed otherwise, ask yourself “Would omitting this item impact the core message the speech is conveying?”. If the answer is “No”, it should be pretty safe to omit it. 🔑 Search keywords: interjections, exclamation marks, filler words, omission, omitting, point d’exclamation ‣

6. Consistency

Consistency means taking decisions on how to tackle specific items and applying the same approach throughout a file. What items should you apply consistency to? Punctuation, capitalization, numbers, dates, spelling, symbols, terminology, interjection rendering... Be sure to stick to these guidelines to ensure the highest level of consistency! The more the consistency, the greater the user experience. 💡 By keeping the consistency, Scribes will be able to increase precision by reducing ambiguity and preventing loss of meaning in case the transcriptions/subtitles are going to be translated into other languages. 💯 Quality tip: You run a search for a proper name and you want to be sure the correct spelling is applied also when files on the same topic arrive in bulk, do you? Or do you opt for one out of several accepted spelling for a single word? Then use the “Glossary” option and add some specific terms there to help grant consistency. Check the section “Terminology” for more details. To check internal consistency in the file you are working on, use the CTRL+F (Command+F) option to navigate through the file and check all instances by typing the first three characters or potential variants. 💡 Still in doubt with some specific items? Post your issue on the French Channel on Discord. Sharing is caring! 🔑 Search keywords: consistency, consistent rules, cohérence, homogénéité, uniformité ‣

7. Special Content ‣

i. Proper Names & Titles Proper names

◦ Capitalize them accordingly. ◦ In case of transliteration from non-Latin languages, check the most common transliteration and use that consistently throughout the file. ◦ Diacritics should be kept in proper nouns and names.

Company/brand/product ◦ Use the correct spelling and capitalization – some of them names might stray from standard conventions. Titles

◦ Run proper searches to transcribe them accurately. ◦ Use italics. ◦ Use sentence case.

💯 Quality tip: The rule of thumb is always to check for spelling company/brand/product names by doing a quick search. Always refer to the official websites to be sure to use the correct names. In the case of book/movie titles, check them on the main online catalogues. If several different spelling/formatting alternatives are available, choose one and be consistent throughout the file. Add it to the Glossary! 🔑 Search keywords: proper nouns, spelling proper nouns, titles, product name, company name, brand name, titres, entreprises, produits, marques, noms propres ‣

ii. Profanity & Curse Words

DON’T censor profanity or curse words. If they appear in the audio, transcribe them word for word. Said: Merde ! Tu m'fais chier avec tes conneries, quoi. DO

DON’T

Merde ! Tu me fais chier avec tes conneries. Tu m'ennuies avec tes bêtises. 💡 Exception: If the curse words are censored and beeped in the audio, DO use asterisks in the transcription/subtitles. 🔑 Search keywords: profanity, curse words, grossièretés, gros mots, insultes ‣

iii. Incorrect Grammar   

DO type the correct spelling in case of mispronunciation of words in the speech. DON’T fix incorrect grammar or words as spoken if the message is not altered. DON’T use the [sic] tag and don’t edit the speech. Audio: “Si j’aurais su, j’aurais pas venu” Audio: “J’adore le poison” (but they meant “poisson”) DO

DON’T

Si j’aurais su, j’aurais pas venu Si j’avais su, je ne serais pas venu J’adore le poisson.

            

J’adore le poison.

💡 Grammar: Focus on flavours For a French speaker not native in frCA, frBE or frCH, note that their grammar can sometimes throw you off. Among other things, you will possibly notice: Modifications in genders (“c’est ma business” 🇨🇦, “faire le party” 🇨🇦, “ils” used instead of “elles” 🇨🇦...) Modifications in numbers (e.g.: “le monde” used just as frFR would use “les gens” → “tous les matins, il y a le monde qui attendent sur le quai.” 🇨🇦) Omissions of the relative pronoun “que” in sentences such as “je pense elle ne viendra pas” 🇨🇦 Usage of "que" as an all-purpose relative pronoun (”le garçon que je t’ai parlé” 🇨🇦) Usage of embedded interrogative pronouns instead of relative pronouns (”je comprends qu’est-ce que tu veux dire” 🇨🇦) Regularization of “aller” (“je va”, “que j’alle, qu’il alle, que vous allez…” 🇨🇦) Lack of the prepositions traditionally associated with certain verbs (”J’ai un enfant à m’occuper” 🇨🇦) Usage of the indicative where frFR would have used the subjunctive (”il faut que tu viens” 🇨🇦) Usage of “vouloir” instead of “aller” for the expression of near future (”il veut pleuvoir” 🇨🇭) Usage of the infinitive instead of the present subjunctive (”Voici des tartines pour toi manger” 🇧🇪) Usage of “savoir” instead of “pouvoir” (”tu sais m’aider, s’il te plait ? 🇧🇪) Modifications in word order (”tu vas ça manger” 🇨🇭) Uses of prepositional accusatives (e.g. usage of the preposition “à” to introduce an object that would be direct in frFR → “aider à quelqu’un” 🇨🇭) DON’T assume these are mistakes and DON’T attempt to correct any of them. To the customer who has requested a specific flavour when uploading their file, even a “de” replacing an “à” or a “dont” replacing a “que” would be considered translation into frFR (and even possibly offensive). 🔑 Search keywords: incorrect grammar, grammar issues, grammar violations, grammaire, corrections grammaticales ‣

iv. Acronyms & Abbreviations  

DON’T hyphenate acronyms (except av. J.-C. and apr. J.-C.) DON’T use periods in abbreviations where not needed. DO

DON’T

Hercule est né vers 460 av. J.-C La SNCF est Hercule est né vers 460 av. JC La S.N.C.F est en grève, le Petit Chaperon Rouge n’a pas pu en grève, le Petit Chaperon Rouge n’a pas pu rendre visite à sa mère-grand. Les BTS sont rendre visite à sa mère-grand. Les B-T-S sont

les rois de la K-pop.

les rois de la K-pop.

💡 Et cetera and its more common abbreviation, etc., are used to show that a list of at least two items is incomplete. Either form can be used. However, choose an option and stick with the same one throughout the file for keeping consistency. 🔑 Search keywords: acronyms, abbreviations, abbreviated words, abréviations, acronymes, sigles ‣

v. Websites & E-mail Addresses  

Websites should be transcribed as they are written normally in the URL. E-mail addresses follow the same rule. DO

DON’T

happyscribe.com

happy scribe dot com

[email protected] scriby at happy scribe dot com m 💡 What if the speaker spells out their e-mail address? Follow the rule for spelt-out items under the section “Capitalization”. 💯 Quality tip: Stick to “e-mail” hyphenated spelling. 🔑 Search keywords: websites, email, e-mail ‣

vi. Onomatopoeias

Since we are currently providing clean transcriptions and subtitles, onomatopoeias should not be included in the text unless they convey a meaningful message, i.e. they are the only answer in a two-person dialogue or they are said by the speakers instead of actual words, for instance. Said

DO

-Je lui avais dit de ne pas grimper. Évidemment, elle ne m’a pas écouté. -Oui, elle est montée et... badaboum. -Exactement. Paf, elle est tombée.

-Je lui avais dit de ne pas grimper. Évidemment, elle ne m’a pas écouté. -Oui, elle est montée et... badaboum. -Exactement. Elle est tombée.

Text in red should not be included (not needed since the meaning is conveyed by “Elle est tombée”), text in green should be (or sentence would be incomplete and meaning lost). DO spell the onomatopoeias as per the list in the link below: Liste non exhaustive d’onomatopées

💡 Check out the “Text Reduction” section for more details on how and when to reduce the text being compliant with our service level. 💯 Quality tip: Onomatopoeias should be in italics. 🔑 Search keywords: onomatopoeia, sounds, atmospherics, onomatopées ‣

vii. Atmospherics & Sounds

Since we are currently providing clean transcriptions and subtitles, atmospherics and sounds should not be included. If the speaker is interrupted by either of them, only the spoken speech should be typed. 🔑 Search keywords: onomatopoeias, sounds, atmospherics, sons, rire, fond sonore, bruit, bruitage, interruptions ‣

viii. Religious References

Use numerals and follow conventionally written formatting for chapter/verse references of religious texts. DO

DON’T

Jean 3 :16 (said : chapitre 3, verset 16 de l'Évangile Jean 3 :16 (only case with no space after selon Jean) a colon) 🔑 Search keywords: religion, religious, bible, citation biblique ‣

8. Special Formatting Italics, Bold and Underline

Happy Scribe Editor supports Italics formatting only. Stick to the chart below for its correct usage. DO use it for

DON’T use it for

◦ Onomatopoeias (only spoken by the speaker) ◦ Voice-over ◦ Proper names ◦ Foreign narration (no documentaries, no reportages) ◦ Unfamiliar foreign loan words which appear in words and phrases which do not appear in the nominated your language’s nominated dictionary for your language ◦ Titles of books, periodicals, dictionary, e.g. e-mail ou works of art, albums, movies, TV shows, radio shows, video week-end for French games, etc. ◦ The speaker is not in the scene(s), not merely offscreen, behind a door or out of shot ◦ Dialogue that is heard through electronic media, such as a phone, television (especially if we see the television and hear the audio), computer, loudspeaker, non-sentient robots, robotic voices or AI, etc ◦ Songs ◦ Metalanguage instances e.g. The word computer is

made of 8 letters 🔑 Search keywords: italics, titles, book, movie, off-screen, when to use italics, italique, titres, livres, films ‣

9. Language-specific Items – French ‣

i. Punctuation ‣

i.a Punctuation – frFR, frBE, frCH 🇫🇷🇧🇪🇨🇭 Punctuation Sign

… Ellipsis

Usage

DOs & DONT’S

◦ Speaker’s trailing off or pausing significantly ◦ Speaker’s cutting off and changing direction

◦ Use a capital letter after ellipsis if the speaker is starting a whole new sentence. ◦ Do not use the ellipsis to introduce a paragraph after a video editing cut (the video is edited in a way that it interrupts the speaker and the next paragraph starts in the middle of a different sentence). A hyphen should be used to mark the interruption. The video cut should be considered as someone else interrupting the speaker and no ellipsis is needed to introduce the next scene.

◦ For vowel elision ◦ For words ‘ Apostrophes voluntarily used in an interrupted form (ex: la mise en prod’)

« Quotation » Marks

◦ Don’t use apostrophes as accents

◦ Capitalize the first word of the quote ◦ If a quote is nested within ◦ Direct speech (spoken, thought or another one, use English quotation written) ◦ Quotes ◦ If the speaker marks (””) for the second one ◦ Don’t mentions “entre guillemets” or is air use quotation marks for titles, unless quoting ◦ The speaker is on the they are part of the title itself ◦ Don’t screen and is reading out loud use quotation marks for metalanguage reference (see italics)

? Question & ◦ Questions ◦ Interjections and shouts ◦ Put them outside the quotation ! Exclamation only marks unless they belong within the Marks quoted text ◦ Always include a hard

space before

◦ Lists ◦ Improving reading flow

◦ Use it before conjunctions (mais, donc, etc.) ◦ Use after quotation marks if sentence is still going

. Periods

◦ Ending sentences

◦ Put it inside the quotation marks ◦ Don’t add another period to names or brands if they end a sentence e.g., I work at Yahoo Inc. NOT Yahoo Inc.. ◦ Don’t use in acronyms

: Colons

◦ Introducing lists or explanations ◦ Introduce direct speech

◦ Include a hard space before and a single space after

; Semicolons

◦ Linking sentences which are closely ◦ Don’t over-use it ◦ Include a hard related ◦ Complex lists with commas space before and a single space after within items

- Hyphens

◦ Compound names ◦ Abrupt interruptions and cutoffs of speakers —no space should be added afterwards ◦ Separating letters in spellings e.g., Happy is spelt H-A-PP-Y ◦ Introducing two speakers in ◦ Don’t use it for line breaks ◦ Never subtitles (no space) ◦ A hyphen use em-dashes for all this should be used to mark the interruption. The video cut should be considered as someone else interrupting the speaker and no ellipsis is needed to introduce the next scene.

— Em-dashes

◦ Marking parenthetical clauses that are grammar-independent in transcriptions e.g., I took my sister’s car without asking her — she never asks when borrowing my clothes anyways — and I drove to town.

, Commas

Single space ◦ Separating parts of the speech 💡 What punctuation to use with interruptions?

◦ Never use en-dashes or commas for this ◦ Add space before and after an em-dash ◦ Don’t use them in subtitles. Rather use commas.

◦ Look for double spaces and delete them ◦ Don’t include any space after hyphen for introducing two speakers

Scenario

Punctuation sign

The speaker interrupts themselves

Ellipsis

The speaker is interrupted by someone Hyphen (no space before) Mirabelle: À propos de ce else or the video is edited in a way that Bruno, je veux- Pepa: Ne parlons pas de Bruno. the speaker is interrupted The speaker is interrupted by someone Hyphens (no spaces before or after) Mirabelle: À else and then resumes with unaltered propos de ce Bruno, je veux- Pepa: Ne parlons pas de grammar Bruno. Mirabelle: -en savoir plus sur ce Bruno. 💡 Air quotations should be written as punctuation signs as well as any mention to “quote unquote”. When you stop “working”, would you mind come and give me a hand with all this mess? 🔑 Search keywords: punctuation, interruptions, spacing rules, space, ponctuation, espace, interruptions, incise, tiret, tiret cadratin, deux-points, point de suspension ‣

i.b Punctuation – frCA specificities 🇨🇦

See below for punctuation signs that are used in a specific way in Canada. For all other punctuation signs, please refer to the previous section as frCA will follow the same rules as frFR, frBE and frCH. Punctuation Sign

Usage

DOs & DONT’S

Question & ? Exclamation ! Marks

◦ Questions ◦ Interjections and shouts only

◦ Put them outside the quotation marks unless they belong within the quoted text ◦ DON’T add a space before

: Colons

◦ Introducing lists or explanations ◦ Include a single space after ◦ ◦ Introduce direct speech DON’T add a space before

; Semicolons

◦ Linking sentences which are closely related ◦ Complex lists with commas within items

◦ Don’t over-use it ◦ DON’T add a space before

🔑 Search keywords: punctuation, interruptions, spacing rules, space, ponctuation, espace, interruptions, incise, tiret, tiret cadratin, deux-points, point de suspension, frCA, français canadien, québécois ‣

ii. Capitalization ‣

When to use upper case Item

💡 Further Explanation

DOs & DONT’S

First word Jane était surprise. Elle dit : « C’est of a Standard sentences Sentences in quotes vous, Tarzan ? » sentence

Proper nouns

People Countries States Regions Cities When a proper name qualifies a geographic common noun Vessels Institutions Genus Ethnic Group

Jean Perrier le Canada la Floride les Flandres Paris la mer Méditerrannée le Titanic l’Union Africaine Homo sapiens les Mayas

Compass Points

When referring to specific geographic l’Allemagne de l’Est est au nord de la areas or part of proper names only France.

Titles

People titles Formal job titles: ◦ referring to specific people ◦ when representing a company

Monsieur, Mme Chef de la direction ; ministre de l’Environnement

‣ When to use ALL-CAPS Capitalized Item

💡 Further Explanation

OST (Onscreen text)

Match OST capitalization and avoid using all-caps in the On-screen text following cases: ◦ Handwritten notes ◦ Excerpts from Forced Narrative books/papers ◦ Social Media messages ◦ Text messages ◦ Long passages of OST

Spelt-out words

When a speaker Happy is spelt H-A-P-P-Y. spells out a word

DOs & DONT’S

‣ When to use Lower Case Capitalized 💡 Further DOs & DONT’S Item Explanation

Titles

Job titles

Le préfet, le curé, l’avocat, le manager... For exceptions, please refer to this page : https://www.lalanguefrancaise.com/articles/guide-complet-usagemajuscules-francais

Variables

Variables

Audio: 3 fois x Typed: 3x

‣ When to use Special Capitalization Capitalized Item

💡 Further Explanation DOs & DONT’S

Titles

Books Movies

Use sentence case

Yahoo! AirPods 💯 Check on official Brand-relevant Company names Product websites/reference material the proper Items names capitalization Acronyms

Institution acronyms Product types Diseases

AfD UNESCO SaaS COVID-19 💯 Always check the proper capitalization on reliable resources

‣

iii. Numbers

‣ 1. General Rules Item

Rule

DO

0-9

Numbers 0-9 are written out.

J’étais bien assez occupé à regarder ses têtes. Il en avait trois.

+10

Numbers above 10 are written in figures.

L’ordre du Phénix compte 23 membres. À Poudlard, il y a 3 500 fenêtres.

Million Billion Trillion and above

The above rules apply. Descriptors are kept.

La saga Harry Potter s'est vendue à plus de 450 millions d'exemplaires. Son adaptation au cinéma a généré sept milliards de dollars.

Starting a sentence

Numbers at the beginning of a sentence are written out regardless of the actual amount or number.

Trente-six élèves attendaient devant la cabane d’Hagrid.

Ordinals

Ordinals 0-9 are written out. Ordinals above 10 are abreviated. la deuxième année la classe de seconde (for proper abreviations, please refer Il est arrivé à la 123ᵉ place. to the official rule)

‣

1.1 Exceptions Item

Rule

DO

Decimals

Always in figures

La valeur appriximative de Pi est 3,14.

Dates

Days and years always in figures

Mon fils est né le 17 mai 2020.

Decades

Always in figures

L’action se passe dans les années 30.

Centuries & millennia

Always in Roman numerals with superscripts

Le XIIᵉ siècle

Measurements

Always in figures Units of measurement are to be spelt out

L’arbre mesurait bien 6,10 mètres de haut. Elle a pédalé 5 kilomètres.

Currencies

General Rules apply. Use currency symbol with figures and spell out currency when numbers are written out

Il a deux euros. Ça coûte 124 €. L’immeuble a été vendu pour trois millions de yens. Cette erreur leur a coûté 25 milliards de dollars.

Addresses

Always in figures

Il habite au 25 rue de la Paix.

Name of a number

Always in figure

Au Japon, on dit que le chiffre 4 porte malheur. Ouvrez vos livres à la page 232.

List of numbers

Always in figures

Ils avaient cinq enfants, âgés de 14, 12, 9, 7 et 4 ans

Combination of numbers

Always in figures

Matilda savait lire à 3 ans, écrivait des poèmes à 5 ans et résolvait des équations différentielles à 12 ans.

Repeated numbers in consecutive sentences/capsules

Always in figures

-J'ai lu que vous aviez 2 500 employés- -3 500. -Excusez-moi? 3 500. Nous avons 3 500 employés.

Religious references

Always in figures

Jean 3 :16 (said : chapitre 3, verset

16 de l'Évangile selon Jean) Phone numbers

Always in figures

Mon numéro de portable est le 06 23 40 00 00

Percentages

Always in figures (with hard space before the symbol)

Dans 99 % des cas, les lamas ne parlent pas.

Range of numbers

Always in figures unless at the Ce centre aéré accueille des enfants beginning of a sentence de 6 à 12 ans.

Time

Always in figures, unless in the case of duration in which case general rules apply

Kings, political regimes Always in Roman numerals & dynasties with superscripts

Je finis le travail à 7h00 J’ai travaillé sept heures. Je travaille 24h/24 et 7j/7. François Iᵉʳ Louix XIV la IIᵈᵉ Republique

‣ 2. Number Separators Item

Punctuation Sign

DO

Decimals

Comma

3,14

Thousands

Hard space

3 000

Dates

/ Slash if all in figures

Phone numbers units -

🇫🇷 🇧🇪🇨🇭Hard space 🇨🇦 Hyphen

11/09/1989 06 50 56 00 00 1-508-555-2232

Prefix

() Rounded parenthesis

(+33) 333 00 00 000

Ranges

- Hyphen

1-12

💯 Quality tip: You found a number in the audio track of the file you are working on, but you don’t remember how to handle that. Quickly surf the Guidelines by using CTRL+F (Command+F) and search for the number type is causing you troubles. 🔑 Search keywords: numbers, decimals, million, date, time, measurements, nombres, chiffres, pourcentages, unités de mesures, date, heure ‣

iv. Date & Time Format Forma Rule t

DO

Date

◦ Follow the format used by the Audio: L’entreprise a été crée en janvier 2017. Le un speaker(s), also in the case of un 2017. Typed: L’entreprise a été crée en janvier centuries 2017. Le 01/01/2017.

Time

◦ Follow the format used by the Audio: Elle se lève à sept heures du matin. Typed: speaker(s) Elle se lève à 7h00 du matin.

🔑 Search keywords: date, time, heure, jour ‣

v. Symbols & Currencies

Special rules apply to symbols and currencies as per the chart below: Item

Rule

DO

Ampersand ◦ Use only if part of a proper name or a Les deux équipes R&D ont travaillé de & common acronym concert. ◦ Use only if included in the speech. ◦ Use the currency symbols with figures Currencies only. ◦ Symbols should go after the number, with a hard space in-between. ◦ Don’t convert.

Ce mois-ci, l’euro a baissé. Je les ai payés 12,99 € pièce. L’erreur leur a coûté plus de 150 000 € à Paris et des millions de dollars outre-Altlantique.

◦ Use the percentage symbols with Percentage figures only. ◦ Symbol should go after % the number, with a hard space inbetween.

J’aime les statistiques et les pourcentages. Par exemple, dans 99 % des cas, les lamas ne parlent pas.

🔑 Search keywords: symbols, percentage, currencies, symboles, pourcentage, monnaie, argent, euro ‣

vi. Mathematical Items

Please stick to the rules below, unless otherwise specified: Item

Rule

DO

Math equations

Use figures for numerals and spell out math operators.

20 plus 4 fois 0.5 égale 22

Greek letters

Spell them out. (unless within formulas)

Alpha, Omega, Pi sont très utilisés en mathématiques

Measurements

Spell them out.

Un angle droit fait 90 degrés Il faut rajouter 1,5 litre de lait. L’insecte mesure 25 millimètres.

Fractions

Spell them out.

Les deux-tiers de mes amis sont partis en vacances.

Trigonometric functions

Spell them out.

Nous allons maintenant calculer le cosinus de cet angle.

Variables

Use symbols and don’t spell out Audio: 3 fois x plus 4 fois y égale 23 the math operator. Typed: 3x plus 4y égale 23

Superscript/Subscript

Use superscripts/subscripts (the Audio: x à la puissance 6 Typed: x⁶ Editor allows its usage). Audio: H2O Typed: H₂O

💡 For subtitles of specific science courses where maths are being written on the blackboard by a teacher, please consider the blackboard as instructions to be followed over the general guidelines. 🔑 Search keywords: numbers, math, mathematical, maths, mathématiques, formules, nombre, fraction, indice, puissance, unités de mesure ‣

vii. References      

Use the following references in case you have any doubts on items not included in these language guidelines: Scribbr: https://www.scribbr.fr/category/elements-linguistiques/ La langue française: https://www.lalanguefrancaise.com/ La ponctuation: https://www.la-ponctuation.com/ Académie française: https://www.academie-francaise.fr/questions-de-langue Office québécois de la langue française 🇨🇦: https://www.oqlf.gouv.qc.ca/accueil.aspx Banque de dépannage linguistique 🇨🇦: http://bdl.oqlf.gouv.qc.ca/bdl/ ‣

10. Terminology ‣

i. Domain Relevance  o 

Always refer to the context to catch the appropriate words or phrases. This is especially important for proper nouns or industry terminology. Lack of proper research might lead to accuracy issues and impact the final quality of your job.

💯 Quality tip: Some foreign words have become part of the common terminology for specific verticals. Depending on the industry, it is not uncommon to see how certain loan words have become part of a specific industry domain. DON’T use tags for loan words or any foreign word that is domain-specific. Example: The English term influencer is part of the specific terminology of Social Media almost across all languages. If the video that you are transcribing/subtitling is about Social Media, then you needn’t use any italics or tags when typing the term influencer. If the video you are transcribing/subtitling is not about Social Media, and the term influencer is used, then it should go in italics. ‣

ii. Glossary 

  

The Editor allows customers and Scribes to add a specific set of terms that should be used in one or multiple files. Why should you stick to the glossary? Because this will: Ensure terminology consistency – internally and across related files. Speed up your task – you won’t need to search elsewhere. Minimize accuracy issues. ‣

iii. On-screen Text   

Stick to the terminology of the on-screen text, unless: The on-screen text has typos. In this case, flag this to the Admins. The glossary provided by the customer includes options that differ from the onscreen text. The instructions state otherwise. 🔑 Search keywords: terms, terminology, glossary, loan word ‣

11. Notation Tags

If the file you are working on has unclear, muffled portions of audio, you can use one of the notation tags listed below, including a timestamp of the audio location. Scenario

Tag

DO

Unintelligible or inaudible [inaudible hh:mm:ss] words are stated.

[inaudible 00:01:32]

Multiple people speaking over each other causes any [diaphonie hh:mm:ss] one of the speakers to become unintelligible.

[diaphonie 00:00:52] Transcribe what is discernible.

Portions of audio in a foreign language.

[langue étrangère hh:mm:ss]

[langue étrangère 01:02:02] Indicate where they begin with a timestamp and the tag “foreign language”. Don’t transcribe non-French audio. If a file is mostly or entirely in a foreign language,

check “Foreign Language” section. The speaker starts singing and it is more than 10 seconds long.     

[chant hh:mm:ss]

[chant 01:59:02] If it takes less than 10 seconds, transcribe the words to the song in italics.

General DON’Ts DON’T leave out the timestamp. [inaudible] DON’T use upper case/all-caps in any notation tags. [Langue Étrangère 02:03:55] DON’T use your own notations. (diphonie 02:03:55) DON’T include the name of the foreign language even though you spot it. [Chinois 02:03:55]

DON’T use [langue étrangère 02:03:55] for loan words. 💡 At Happy Scribe we currently DON’T use any [guess] tag. When you are really struggling with some words, it’s safer to use the [inaudible hh:mm:ss] tag. 💯 Quality tip: Do you want to use a tag accurately? Type [ on the Editor at the exact point you want to add a tag and choose from the drop-down menu. Windows shortcut: Alt+0091 Mac shortcut: Alt+Shift+ ^ key Punctuation and Tags When the tag is replacing a word, a phrase or a portion of a sentence, DO use standard punctuation afterwards. However, when a tag is used to replace a whole sentence or paragraph, DON’T include any punctuation sign after the tag. 🔑 Search keywords: foreign, foreign language, inaudible, crosstalk, singing, notation tag, loan word, langue étrangère, diaphonie, chant, chanson ‣

12. Transcription Specifics ‣

i. Multiple Speakers

In the case of multiple speakers, a labelling strategy should be followed. Refrain from creating your own descriptive labels and rather stick to the following rules based on different scenarios: Scenario

Rule

DO

The speaker’s name can be reasonably inferred from the audio/video.

Use speaker’s name label

Jean Perrier Docteur Carrère

The speaker’s name cannot be reasonably inferred from the audio/video, but the job title or role can be.

Use job title or role labels

Intervieweur/euse Professeur Étudiant/e

The speaker’s name cannot be

Use sequentially numbered labels Orateur 1 Orateur 2

reasonably inferred from the audio/video, nor can be the job title or role. There are too many speakers to consistently track who says what (e.g. classroom discussion, focus group).

Use group labels. Don’t use it as a substitute for reasonable speaker Étudiants Audience identification. Don’t include any Équipe de tournage number.

💡 What to do in case of crosstalk? Check the “Notation Tag” section. What to do in case of interruptions? Check the “Punctuation” section. 💯 Quality tip: Remember you must assign a speaker to all paragraphs, even if there are multiple consecutive paragraphs from the same speaker. If the client provided the list of speakers, be sure to stick to it. 🔑 Search keywords: multiple speakers, various speakers, plusieurs intervenants, intervenants divers ‣

ii. Segmentation  

Avoid writing excessively long paragraphs, even if there is only one speaker in the file. Look for natural points in the speech, such as a shift in topic, emphasis, or reference, to start a new paragraph. 💡 How to assess whether a paragraph is “too long"? Check this out. 💯 Long paragraphs are now warned by the quality check tool! 🔑 Search keywords: long paragraphs, longs paragraphes ‣

13. Subtitles Specifics ‣

i. Multiple Speakers  

When subtitling, DON’T add any speaker label. Use hyphens (not followed by any space) to separate speech uttered by two speakers speaking at the same time, or speaking too closely to each other – putting them in one capsule preserves CPS limits and avoids fast transition from one capsule to another.



Only one speaker per line is allowed. DO

DON’T

-Bonjour. -Salut. Tu vas mieux ?

-Bonjour. - Salut. Tu vas mieux ?

🔑 Search keywords: multiple speakers, various speakers, discours simultanés, prises de paroles simultanées, chevauchements ‣

ii. Segmentation  

‣ 1. Line Treatment Subtitles can have at most two lines, even in the case of multiple speakers. Try to always stick to 1 line unless the character per line (CPL) limit is exceeded. DO

DON’T

Elle est allée à Paris le weekend dernier. (42 CPL)

Elle est allée à Paris (22 CPL) le weekend dernier. (19 CPL)

‣ 2. Line Breaks Ideally, try to seek subtitles balance by keeping the first line shorter than the second line. However, remember that logical line breaks override subtitles balance.

Lines should be broken at logical points, matching a punctuation sign like a period, comma or colon whenever possible. If the break has to be elsewhere in the sentence, avoid splitting the parts of the speech in the chart below. DON’T sacrifice logical line breaks to keep a pyramidal text distribution in subtitles.

  

  

Parts of the speech

DO

DON’T

Noun

Ils ont organisé une conférence à Rome.

Ils ont organisé une conférence à Rome.

Subject/Pronoun and verb

Ils ont organisé une conférence à Rome.

Ils ont organisé une conférence à Rome.

Parts of a complex verb

Ils ont organisé une conférence à Rome.

Ils ont organisé une conférence à Rome.

Article and noun

Ils ont organisé une conférence à Rome.

Ils ont organisé une conférence à Rome.

Preposition and following phrase

Ils ont organisé une conférence à Rome.

Ils ont organisé une conférence à Rome.

Conjunction and following item

Ils ont organisé une conférence à Rome, mais je n’y suis pas allé.

Ils ont organisé une conférence mais je n'y suis pas allé.

💯 Quality tip: Are you still struggling to understand what a logical point is? Have a look at the sentences below for further examples. The ideal break places are marked by green slashes. The orange slashes indicate where it would be OK to break if breaking at the green slashes were not possible. C'est un morceau/de prose,/très long/et verbeux,/que personne/ne connaît,/et dont personne/ne va se souvenir. Jean-Pierre souhaite se rendre/au supermarché,/mais, à ce qu'il me semble,/tous les magasins sont fermés/durant la Journée/de la Traduction. Je me suis réveillée,/en plein décalage horaire,/à trois heures du matin,/dans mon lit,/et j'ai tout de suite appelé/Anne-Marie Dupont/afin de lui parler/de mon interview. 🔑 Search keywords: line break, number of lines, découpage ‣ 3. CPL The subtitles CPL should not exceed 42 CPL. Whenever exceeded, it is marked in red in the Editor. If the line is over 42 CPL, you can either: Break the subtitle into two lines by pressing Shift+Enter; Split into multiple subtitles by pressing Enter to stay within this limit; Omit content without altering the core meaning of the speech. Check the section “Text Reduction” for further details.



Delete any blank space at the end of each line.

💯 Quality tip: Did you have to break many lines to meet the CPL and you are afraid of having messed it up with line-break rules? Be sure that no line ends with an article or a preposition and watch out for those ending with an auxiliary verb – they might potentially contain issues. Remember to check the time sync as well. 💡 CPL for portrait mode is 32. For further details, check the section “Portrait Mode”. 🔑 Search keywords: line break, cpl, character limit, saut de ligne, limite de caractères ‣

iii. Time Sync 



‣ 1. CPS The subtitle CPS (Characters per second) is an indicator of the comfort of the viewer. A high CPS indicates that the viewer will have to read faster, making the experience less comfortable. Values higher than 20 CPS indicate a degraded viewer experience, so you should make an effort to try to reduce it when possible. Whenever exceeded, it is marked in red in the Editor.



 

There is a slight tolerance in exceeding the 20 CPS soft limit if the speaker talks too fast and it is not possible to reduce the subtitle further without losing meaning or worsening the final user’s experience. 💡 In case of content for children, the CPS drops down to 15. ‣ 2. Sync To be sure your text is synced with the audio and thus granting a great experience to the final user, stick to the guidelines listed below (in order of priority): Subtitle appearance should coincide with speech onset. Subtitle disappearance should coincide roughly with the end of the corresponding speech segment. Your subtitles should never anticipate the speech, so please avoid stretching the subtitle event to the left. If needs be, you can slightly stretch the event to the right. Remember that when stretching the capsule to the right, it will be hanging up on the screen, so avoid to leave it there for too long not to worsen the final users’ experience. The end of the capsule should match the end of the speech as much as possible. The subtitles cannot stay on the screen more than 1.5 seconds after the speech has stopped under any circumstances.





Avoid creating subtitles that straddle a shot change (i.e. a subtitle that starts in the middle of shot one and ends in the middle of shot two). You might need to split a sentence at an appropriate point or delay the start of a new one to coincide with the shot change. To meet the 20 CPS limit, you can split or merge subtitles or slightly extend their timespan. 💡 Use the timeline interface located at the bottom of the editor to synchronize the subtitles with the audio and extend or shorten their timespan. 💯 Quality tip: While you should aim to meet the CPS, it is paramount that the text on the screen coincides with the speech – the final user’s experience is worsened when they have to read something different from what they are hearing. Avoid pushing the text forward or backwards to meet the CPS if this leads to a mismatch between speech and text. In the example below, for instance, the third subtitle has been pushed back to meet the 20-character limit of the second one (boxes in green show where the synced subtitles should be appearing). It means that a new sentence in the speech would start while the previous subtitle is still on screen, this worsening the final user’s experience, even though the CPS has been met. In general, always try to avoid stretching the subtitle event to the left as this might lead to a mismatch between the start of the text and the start of the speech.

🔑 Search keywords: cps, time sync, sync ‣

iv. Portrait Mode and Square Videos

  

Portrait Mode In some cases, you might find the video has been recorded vertically. This can be easily identified because the picture will not cover the entire screen. Please follow the adjusted criteria (in order of priority): Maximum 32 CPL Keep 20 CPS Ignore the number of lines appearing on the video, the font size will be adjusted later.



Square Videos In some cases, you might find the video has a square format (1:1). Please follow the adjusted criteria (in order of priority): Try to stick to 36/38 CPL if possible 💡 How to adjust the criteria? 1. Open the subtitle Editor 2. Click on “Edit subtitle limits” 3. Adjust the CPL accordingly

🔑

Search keywords: vertical video, portrait mode, video format, square video, mode portrait, vidéo verticale, format vidéo, format carré 💡 How to prioritise subtitle specifics?

‣

14. Useful Tools    o 

Here is a list of life-saving tools for transcriptionists and subtitlers: Grammarly Language Tool Sound booster Chrome extension Background noise-cancelling extension ‣

15. Reminders    

 

Always check if there are specific instructions from the customer for a file. In that case, the customer’s instructions prevail over these Guidelines. Never transcribe in an external editor (Word for example) and then paste the text into Happy Scribe's editor. Be sure to run a spell-check by using Grammarly or any other available extension available for your browser. If you struggle with some specific words or a part of the speech, be sure to run some searches on the Internet. Check pages related to the topic of the file you are working on for better results. If you cannot find the answer, reach out to the Tribe on Discord and seek further support. Please note, once you have submitted a file, you cannot alter and correct the text afterwards. If you accidentally submit a file, please reach out to the Admins. Be sure you have the spell-check option/extension/plug-in enabled on your browser. ‣

16. Version Log Global updates Date

Update

3 August 2022

Clarifications on speakers reading out loud

23 August 2022

Prioritisation scale

20 September 2022

Usage of hyphen and ellipsis in interruption due to video editing