Please find below our Golden Rules. These are the most important things for you to remember while you’re typing, to ensure that every file you’re producing is accurate and consistent.
- Word for word
You must capture everything that is being said in intelligible sentences that flow and are sufficiently punctuated. But no need to transcribe the ums and errs!
Don’t let sentences run on for too long, and leave out false starts and repeated words that don't add anything of meaning. For example:
Would you-, am I right in understanding that you would go there?
Am I right in understanding that you would go there?
2. Sort of, you know, like
You should leave some of these filler words in the transcript, because they're important for reflecting the flow of the person's speech. Omit some here and there if they are too liberally used by the speaker (e.g. more than once in the same sentence) and are making sentences difficult to read. They should be in between commas, for example:
I like, you know, the way he says his name.
Do you think, sort of, you'd like that car yourself, or do you, you know, think you'd prefer the more expensive one?
3. Spelling (particularly brand names, place names and specialist terminology)
Please use UK spellings throughout. It’s also really important to research brand names, place names and any terminology used - a quick Google search can often find the right term, even if your guess at spelling it is really far off. If the name is listed by the company as having all capitals and it is not an acronym, it should be changed to Title Case. Any special punctuation or formatting should also be amended or removed to fit proper grammatical styles, e.g. Lego, Lovefilm, BBC.
4. Inaudible speech and silence
Sometimes it might not be possible to decipher what the speaker is saying and you will need to make a note of this to the client with an inaudible marker and a timecode that refers to that section of the audio. In the vast majority of audio files, these should be few and far between. The timecodes we use are:
Inaudible phrase: (inaudible 00.04)
Inaudible sentence: (inaudible 00.05-00.10)
Missed word: (? 00.03)
Crosstalk: (Talking over each other 00.30)
If there are any other events which are meaningful to the transcript, you should timecode them like this:
(Moderator leaves room 55.06-59.15)
There should be no need to use colons, semi-colons, dashes, ellipses, brackets or exclamation marks in your transcripts. Quotes should be in single quotation marks.
6. Ten-minute timecodes
Throughout the transcript you’ll need to include markers every ten minutes, so that the client can easily see which part of the audio they’re reading. These should be as near to every ten-minute mark as possible. For example:
What did (TC: 00:10:00) you think of the way this was portrayed?
7. Grammar errors and word changes
It's up to you to ensure files are grammatically correct - even if the speakers aren't! There are also a few minor word amendments you are permitted to make:
yeah/yep = yes
nope/nah = no
gonna = going to
gotta = got to
wanna = want to
should of = should have
’cause = because
8. Identifying moderators and respondents
If there’s a clear person who is asking the questions then we refer to them as the Moderator and their speech is written in bold. The person or people answering the questions are referred to as the respondents and are typed in non-bold.
If a respondent is identified by name in a file consisting of only one or two people, then they must be identified by this name, like so:
Could you start by telling me your name and occupation?
Thomas: Sure. My name’s Thomas and I work in finance.
If you have a respondent's full name you identify them by this the first time they speak and then by their initials thereon in. If you don’t have any part of their name at all you would use M1, M2, F1 or F2 to differentiate between the speakers.
If there’s more than one moderator put the main moderator in bold and any secondary moderators in bold italics.
Speaker ID for 4+ speaker files
With 4 or more speakers the moderator is still typed in bold and respondents in non-bold, but throughout the transcript you ID the respondents with M: for male and F: for female. Like this:
Could you all start by telling me your name and occupation?
M: Sure. My name’s Thomas and I work in finance.
F: I’m Lisa and I’m a teacher.
If it’s a whole group of respondents of the same gender you don’t need to include any ID.
When a client has specifically requested for speakers to be identified throughout you’ll see the words ‘Speaker ID’ underneath the service in the Portal. When this has been chosen you’ll need to do your best to ID speakers either via name, if you have it, or using M1, M2, F1, etc. as IDs.
When you’re typing a focus group you only need to start typing from when the respondents start introducing themselves - there’s no need to include the moderator’s introduction.
9. Translator files
In translator files, it will not be possible to identify the genders of the speakers, as it will be just one person translating a whole interview/focus group from another language. However you must use your common sense to identify when different speakers are talking, and start a new line for each new speaker accordingly, as per our usual formatting.
Numbers from one to twenty should be typed as words and then digits used from 21 upwards. If the number is below 21 but is a date, time, monetary amount or part of a brand name, then a digit can be used, e.g. 16th December, 7:00, £3.60, Channel 4.
11. Split Files
If you’re assigned the first part of a split file, once you have completed the first few pages (after the moderator introduction/respondents have been identified) please fill out the transcriber comments, noting the number of moderators and the number/gender of the respondents. You should then upload this to the portal so that other transcribers working on the file can refer to them and ensure speaker IDs are consistent throughout. If the file needs to have Speaker ID, please note the speaker names in the comments section of the portal.
Please also type a few sentences before the beginning of your section and when you reach the end of your assigned section of audio always end with a full sentence, even if it means going a few seconds past your allotted time. It makes it a lot easier for us to match up the separate parts!
12. Adverts played to the group
Where something out of context e.g. an advert or recording, etc. is played or read to respondents, we do not need to type up the whole thing as the client already has this information. However, we do need to be able to make it clear what is being referred to. Leave a blank line and type the first couple of sentences in regular italics (not bold) on a line of its own. Then put (Advert/Video played/read to group 00.00-00.00) in regular text (not italics) so that the client knows what was played or read out, and from where to pick up the audio again.
Last but not least: if in doubt, use your common sense!