Transcription Instructions


.

_____________ Transcription Instructions ______________

Introduction

Transcription tasks involve listening to an audio file which we call the source file and typing in writing exactly what is said into a document, following a specific set of transcription guidelines.

The task can be done using many different tools, such as:

  • Otranscribe: an online transcription tool
  • Audio player and Word document

General Transcription Guidelines

When transcribing, there are several key areas you need to consider to ensure you produce a quality transcription. Unless a different set of transcription guidelines are provided by your Project Officer, please follow the instructions below.

Technical Guidelines

Guideline Description
Listen Listen to the whole audio clip before you start transcribing. This is important for determining if the audio is of decent quality to transcribe. If it is not (for instance if most of the audio clip is indecipherable due to noise), please inform your Project Officer.
Speaker Tags There may be multiple speakers in the audio files. To mark each speaker throughout the audio file:
  • If speakers are identified by their names at the start of an audio clip, you can give speaker labels. Your Project Officer may also provide labels beforehand. Otherwise, assign them a letter
  • Start each speaker with a new line.
For example:
  • A: Good morning, you’re live on air. What’s your name?
  • B: Hi, I’m John, and I’m with my brother.
  • C: Hi there, I’m Henry.
  • A: Fantastic, good to meet you two.
  • C: Good to meet you too.
  • B: We’re big fans of the show.
Tagging issues within audio segments Audio containing any of the issues listed below should be tagged as indicated.

Please do not transcribe parts that are unintelligible, instead note the time codes for unintelligible sections and tag the issue as advised below.

  1. [issue.noise] - Extremely noisy audio: audio that is too noisy to be transcribed reliably. This could include radio, TV or other noise in the background that does not allow one to understand the speaker.
  2. [issue.speech] - Unintelligible speech: If parts of the audio file has speech that cannot be understood or transcribed reliably (could be because the speaker is speaking too softly, mumbling or multiple speakers are speaking over each other such that you cannot decipher what is being said)
  3. [?] - If there is just a word that is unintelligible, use the tag [?] at the appropriate place in the transcribed sentence.
  4. [issue.foreign] - Foreign speech: if parts of the audio have foreign words other than in English or the specific language for which this task is assigned.
  5. [issue.nospeech] - No speech: audio containing no speech to transcribe (such as audio that is entirely music, background noise, or silence).
Foreign language exceptions The following foreign speech should be transcribed:
  • Names: Write the name as it is.
  • Commonly loaned words: If the foreign language words are commonly used in the source language of the transcription task, then transcribe them as it is.
  • English: If the source language words have been mixed with English words, then just write them as they are - English words in English and the source language words in the source language.
Sounds
  • Don't write anything for sounds such as person breathing or silent gaps between words.
  • If you can hear music or other ambient sounds in the gaps between the words, no need to write anything for that.
Time coding If required by the project, please insert time codes to your transcription.
  • Format: there are many different timestamp formats (HH:MM / HH:MM:SS / HH:MM:SS:MMM). Check with your Project Officer which one is required.
  • Speaker change timestamps: inserted every time a speaker starts speaking. This is the most frequent type of timestamps used.

Linguistic Guidelines

Rule of thumb: never paraphrase or reconstruct what you are hearing and transcribe verbatim (not even to correct grammatical errors by the speakers!)

Guideline Description
Verbatim transcription
  • Always transcribe word-for-word. Write exactly what is said in the same language that was spoken. No paraphrasing, no translating, no comments.
  • E.g. If the speaker says ‘I don’t eat cheese’, please don’t expand that to ‘I do not eat cheese’; write exactly how it is said in the audio.
Spelling and capitalization Spell words carefully and correctly, refer to a dictionary, newspapers, other reference material or colleagues if unsure of a spelling.
  • Capitalize the first letter when writing proper nouns (names of individuals, places etc.) and as per the norm in your language.
  • Check with your Project Officer if there is a style guide available for your language.
Grammar, word-order, speaking errors Do not correct the grammar, word-order, or other speaking errors such as repetition of words.
  • If the speaker says “I think I is going go to to the m* oon” then write it exactly like that.
  • If the speaker repeats themselves, please transcribe the repetition too.
Mispronunciations For mispronounced words, type the word that you believe the speaker wanted to say. For example, if the speaker says ‘argooment’ (meaning ‘argument’), you would type ‘argument’.
  • Do not change the spelling of a word to fit the pronunciation or accent of the speaker.
  • Do pay attention to determine whether it is a slang word and not a mispronunciation. In case it’s a slang word, please transcribe it as it is.
  • Also don’t expand slang words such as wanna, gonna and transcribe them as they are spoken.
Filler pauses Filler pauses include all spoken sounds similar to the examples below, as well as stuttering and stammering sounds: “um”, “uh”, ”err”, “nah”, “eh”, “huh”, “hm”, “mmm”, “ah” Do not transcribe these.
Numbers Spell out numbers
  • e.g. zero, ten, one hundred and fifty, two thousand and twenty etc.
Dates and times Spell out dates and times.
  • Write them exactly as the speaker says, e.g. First of January, two thousand and twenty at eight pm or January the first, two thousand and twenty at eight in the evening.
Punctuation
  • Do not assume punctuations such as exclamation marks and just use a full stop.
  • IIf the speech makes it clear that a sentence is a question then use the question mark.
Partial words When someone says only part of a word, write down the part that you heard. Do not complete the word on your own.
Acronyms Write them as you hear them in the audio file. Do not expand to their full form if the acronym was used in the audio
Symbols Do not use any symbols or currencies (i.e. R,$,@,&,+,-). Spell out the word that was said.
Examples
  • $ write like this → dollar
  • @ write like this → at
  • + write like this → plus

Instructions

If you have any questions or need anything, please do not hesitate to contact your Project Officer. They will be happy to help out!

Request

You will be provided with:

  • Link to the transcription task
  • Audio to be transcribed
  • Project-specific instructions
    • Partner-specific transcription guidelines or requests
    • Glossary

Process

  1. Claiming the task:
    make sure you have claimed the transcription task before beginning any work on it. You can begin once you receive a confirmation email with the task instructions.
  2. Technical preparation
    • Transcription Guidelines: familiarize yourself with the General Transcription Guidelines in this document and with any project-specific guidance provided.
    • Audio Quality: listen carefully to a few samples of the audio to make sure it is of good enough quality. If not, please flag to your Project Officer immediately.
    • Speaker list: if provided, familiarize yourself with the list of speakers to make sure they are correctly identified in the transcription
    • Glossary: if provided, familiarize yourself with the project glossary.
  3. Content inspection: familiarize yourself with the audio file, i.e listen first to understand it and how it flows to get a feel of the concept or story in the source file.
  4. Transcription Process
    • Open the transcription tool and begin transcribing the file. Rewind the audio as many times as you need to make an informed decision for your transcription.
    • If any issues come up, consult with your Project Officer promptly so they can confirm with the Partner, if necessary
  5. Review Process
    • Final listen: listen the audio one final time while reading your transcription, to make sure it is accurate and precise.
    • Spellchecker: run the spell checker to correct any remaining spelling issues.
    • Timestamp adjustment: listen to the audio one final time to make sure the transcription time codes are aligned and precise. This is very important, so please do not skip this step.
  6. Submission: deliver the transcription to the Project Officer. If there are any pending issues or points you would like to highlight, please include them in a short handover message.
    • Your comments are encouraged and valuable, please do share!

Tips

Item Description
Media Control You can speed up or slow down the delivery of your audio or video to facilitate your work.

Keyboard Shortcuts Continuously shifting your hand from your keyboard to your mouse can be time-consuming. Keyboard shortcuts can help you be much more productive! The main ones to try to use are:
  • Play and Pause
  • Rewind or Fast-forward
  • Slow down or speed up
  • Insert timestamp
Different tools have different shortcuts:
Exports You may wish to export your transcription to use your local spellchecker, for example.
You are welcome to do so, but please make sure you update the final version in the transcription system. Project managers need this to be able to deliver the project.
Review process We recommend always doing at least two separate reviews: one for linguistic aspects and another one for technical aspects.
Questions Please, ask questions! Project Officers will be happy to follow up and check with Partners.
This is a vital step in delivering quality transcriptions.

Pre-delivery checklist

  • Confirm that the complete audio file has been transcribed.
  • Perform the linguistic review (accuracy, spelling, grammar, etc.)
  • Perform the technical review (time coding, speaker tags, issue tags, etc.)
1 Like