Professional Documents
Culture Documents
Don't italicize; we have used italics here solely to highlight what is right and wrong in the
example
● Omit unnecessary filler words, for readability
● Omit false starts, stutters, repetitions, etc
● Omit interjections, unless a direct response
● Never omit meaningful words or entire sentences
● Never omit or correct expletives
● Never correct the speaker's grammar*
● Never paraphrase
● Never substitute words
● Never rearrange the order of speech
● Save the file as a Word document
specific terminology
● Google proper spelling of names and brands
● If answer not easily findable, use a common or
phonetic spelling and mark with timestamp
● Spell words consistently
Speaker Identification
The speaker ID is aligned with the left edge of the margin and all other text is indented by one
inch (1”) using a hanging indent.
Lines are single-spaced with a blank line inserted between speakers or paragraphs. The font is
Calibri 12 pt. Use the Word
template we provide to ensure you have the margins set correctly.
● Don't use tab instead of a hanging indent. This makes editing documents during the review
process much more lengthy and time consuming for the quality assurance agent. Tab should
only be used after the colon at the end of a labeled speaker.
● Don't use page numbers, headers, or footers.
● Don't mark the end of the file with “End of Audio” or the duration.
● Set your margins to one inch (1”)
● Set up a one inch (1”) hanging indent
● Ensure font type is set to Calibri, 12 pt, black.
● Use single spacing, with a full line between speakers and paragraphs.
● Keep paragraphs to 8-10 lines, and look for natural points in the speech (a shift in topic,
emphasis, or reference) to start a new paragraph.
● Be sure you understand and implement any customer requests or special instructions.
Brackets:
Notation on the transcript, used as a comment by the transcriptionist. Should not be used as
actual spoken text. For example:
Speaker 1: Sometimes I think it could have been expanded, it was never my intention [inaudible
00:45:08] would be good.
Commas:
Used in a pair in the middle of a sentence to set off clauses, phrases, and words that are not
essential to the meaning of the
sentence. They are not used to indicate a pause.
Diacritical Marks:
These marks that aid in pronunciation are nice to include but are not required. For example:
fiance vs. fiancé
uber vs. über
Ellipses:
Use ellipses to indicate a speaker has trailed off or paused in the middle of a statement, or has
been interrupted by another
speaker. Please treat ellipses like a word with a space before and after. For example:
Speaker 1: John wanted to say he cared but … I guess he could not do it.
Speaker 2: Do you think I could …?
Speaker 1: … he didn’t have the mental stamina.
Hyphens:
Hyphenate two or more words that precede and modify a noun as a unit, especially if the words
include a past participle, a present
participle, a single letter, or a number. For example:
line-by-line scrolling
read-only memory
Note that hyphens are not used to indicate pauses or changes in mid-sentence.
Quotation Marks:
Use quotation marks when a quote is directly stated or implied. The quote can be of themselves
or others. For example:
Speaker 1: Arnold said, “Annie, um, that’s a joke.” Can you believe it?
Speaker 1: Then I told him, “I don’t think I can,” and he said, “Okay then.”
Speaker 1: I said, "What is going on?" Then Sarah goes, "I don't know. Maybe we should find
out." Of course, no one ever told us. We still don't know.
Tagging
[song] You should not transcribe song lyrics.
[foreign language hh:mm:ss]
For any non-English portions of audio, please indicate where they begin with a timestamp and
either the name of the language (if known) or simply ‘foreign language’. If at the beginning of a
sentence, it is capitalized [Foreign language hh:mm:ss], otherwise it is written with the
lowercase [foreign language hh:mm:ss].
[inaudible hh:mm:ss] Used when unintelligible or inaudible words are stated. If at the beginning
of a sentence, it is capitalized [Inaudible hh:mm:ss], otherwise it is written with the lowercase
[inaudible hh:mm:ss].
[crosstalk hh:mm:ss] Used when multiple people speaking over each other causes the audio to
become unintelligible. These should be labeled within the text as they occur. If at the beginning
of a sentence, it is capitalized
[Crosstalk hh:mm:ss], otherwise it is written with the lowercase [crosstalk hh:mm:ss].
['phonetic' hh:mm:ss] When you are unsure of the spelling of a word or name, include your
attempted spelling in brackets with the timestamp. This allows the reviewer to easily double
check the audio during review. For example:
Speaker 1: He is totally a [brony 01:04:48]. Watches it every day.
[hh:mm:ss] Standard format for timestamps indicating the hour, minute and seconds at which
the speech is heard.
Please do not include milliseconds. See below for more information.
Special Tags
Please pay attention to the use of parentheses () vs. brackets [] for each tag type.
Here we demonstrate how to use ellipses, crosstalk, and inaudibles in broken conversation:
Speaker 2: 7th Street and Balfour? That's right where [inaudible 00:00:13] ...
Speaker 1: Yes, 7th Street. [Crosstalk 00:00:34]
Speaker 2: Let's go off the record and adjourn the meeting.
General Timestamps
Use punctuation appropriately to provide a readable document, avoiding run-on sentences and
long-winded paragraphs when you can. If you are certain about affirmative or negative
agreement by, based on context and intonation of the speaker, then add the tag
(affirmative) or (negative); if you are uncertain, use a neutral.
● “um”, “ hmm”, and “mm” are good neutrals to use
● “mm-hmm (affirmative)” vs. “mm-mm (negative)”
● “uh-huh (affirmative)” vs. “uh-uh (negative)” or “uh-oh (negative)”
Interjections:
From other speakers should be omitted when they are simply responding - like “yeah”, “yup”,
and “hmm” - rather than answering a direct question.
Expletives:
Such as “damn” are treated as any other word and transcribed as-is. We do not censor the
transcriptions.
Omitting Words
We strive to accurately capture the meaning and intent of the audio. To that end, we generally
do not correct for grammar, contractions, or slang. For example, you should not change "don't"
to "do not" or "yeah" to "yes".