Transcription top tips for getting a written record
How to get a transcription of your meeting or event
There are many different ways you can get a written record of a meeting, interview, video or event in English or other language and transcription is a popular choice. Because of the time involved in producing them transcripts can, however, be expensive in high volume so here are our top tips for choosing the most relevant service:
Talk with a Language Service Provider (LSP)
Tip 1) Talk with your Language Service Provider (LSP) before your event to decide together if transcription is the best solution for your setting and purpose. Sometimes written summaries of audio recordings or interpreters rendering a live event into English followed by English transcription may be cheaper and quicker than full transcription or translated transcription. Your LSP will weigh up the costs involved and the level of detail you need.
Single step process
Tip 2) Try to have one step where possible. Where it is necessary to translate audio, specialist linguists can, for example, produce translated subtitles directly from audio, or produce summaries to a particular brief. These services may be more cost-effective. When clients do require a transcript plus a translation of that transcript into another language or transcription plus subtitling, the LSP will be happy to do this too.
Clear Audio
Tip 3) Make sure the audio is clear from all speakers (and ask everyone to speak with the recording in mind). This will reduce transcription costs as it simply takes less time to transcribe a clear audio than one which needs the sound adjusting and several replays to transcribe accurately. If you only need transcription into the same language, clear audio may give you a decent result with auto-transcription software – see tip 5!
Clear instructions
Tip 4) Give your transcriber instructions. Which of the following would you like in your transcript?
a. Time codes – Do you need these in order to navigate the audio later?
b. Speaker names – Would you like the speakers noted in full, by their initials or by role?
c. Sounds on the audio – Should background noise, interruptions, laughing, etc. be noted?
d. Silences – Should lengthy silences be noted?
e. Inaudible sections – Do you prefer the transcriber use their best guess where possible or always mark it as ‘inaudible’, with the time code if wished?
f. Verbatim or intelligent verbatim? The latter removes unnecessary fillers and repetition.
Use Automation where possible
Tip 5) Automatic transcription. Every minute of audio takes a human transcriber between 4 and 6 minutes to transcribe, depending on how many elements you’ve included from tip 4. It’s really easy then to see the advantage of automatic transcription software! Auto-transcription can be used if you: need your transcript in the same language as the audio; it’s for informal use; you have clear speakers with standard accents; you don’t need many added notes or formatting. (It can actually take longer to heavily annotate an auto-generated transcript than transcribe manually from scratch.)
Try Trint or Transcribe by Wreally. These will both produce a transcript in the same language as your audio but which may need a few corrections. Trint inserts time codes, ‘pins’ the audio to the generated transcript for easy searching and can be programmed to ‘learn’ certain words. Transcribe by Wreally is great if you want to make edits to an auto-transcript or to transcribe manually as it has high playback precision and speed control.
Note: These software cannot translate audio into other languages. Normally 1 minute of audio takes 10 minutes to translate manually. If the audio includes a lesser-known language or accent, or your audio is in one language and you would like your transcript in another, you will need a human transcriber. 😊
Good luck and we’re happy to answer your questions!
Photo accredited to Kelly Sikkema on https://unsplash.com/
