Transcription Process

One of my projects here with the South Dakota Oral History Project (SDOHP) of the I.D. Weeks Library is creating oral history transcripts. We have thousands of digitized interviews, and with the rise of awareness concerning accessibility, we are on a path to transcribing those interviews. This will not only help with accessibility, but also with citations since the syntax has already been edited for consumption. 

It is important for me to have a process in everything I do, so in hopes that others will begin interviewing, recording, and transcribing, I am going to share my personal transcription process. Currently, we use OtterAI to create an initial transcription, though much editing must be done to preserve meaning, voice, and general cohesiveness. At a minimum, I listen to the interview an even six times. Each step describes what I do during that listen-through.

1. Properly Identify Speakers

   1. Become Familiar with their Individualized Speech Patterns

OtterAI isn’t perfect, and one of the areas it exposes its imperfections is in identifying speakers. As you listen, you will become familiar not only with the speakers’ voices, but also with how they speak. For example, in SDOHP1491, Emil Jerke (the interviewee) is bilingual. More than this, Jerke is a storyteller who focuses on small examples to give the listener a bigger picture. He uses asides to give context and works through memory and language to get to the heart of what he means.

1. Correct Words and Fix Minor Grammatical Errors

   1. Highlight Unclear Words

It would be quite difficult to correct all grammatical errors on the first or second listen-through so it is important to focus on what can be fixed quickly here. Rather than trying to insert all asides and false starts, fix capitalization and end punctuation errors. Essentially, fix what grammatical errors you can without stopping the tape. Doing this can help processing go a little faster since you aren’t looking at making every correction.

1. Correct All Grammatical Errors

   1. Attempt to Decipher Unclear Words

   2. Highlight (in a different color) Grammatically Complex Sections

This listen-through should take the longest. By now, you should be familiar with speech patterns, speaker’s usage of false starts and asides, what they are trying to say, etc. You will need to pause and rewind quite often during this time. Now that you are more familiar with the topic, it is easier to conduct research and find out what those previously unclear words are. If it is still unclear, you can mark it in brackets. Some sections may be complex grammatically, especially considering this is a work reflecting spoken word. If you find yourself going back and forth on how to edit it, highlight it and move on. Clarity often comes when you are able to focus on those fewer, more complex sections than on the entirety.

1. Make Final Corrections

   1. Final corrections include anything regarding speakers, grammar, unclear words, and more.

This is a dizzying time. You’ve gone through the document several times already, and yet you are still just over halfway done. Yet, it is important that at this stage, you make the transcription as clean as possible. It would seem this would be the easiest stage, but you have left much of the bigger puzzles left to solve for now. While there tends to be far fewer pauses when making corrections on the fourth listen-through, the pauses will likely be longer. This is good; it means you are paying attention and taking care.

1. Ensure Formatting is Correct

   1. Give unclear sections a final review and attempt to decipher.

   2. Make sure all grammar reads well and accurately (accurate, in terms of how the speaker speaks).

This is it, the moment you’ve been waiting for: to make sure what you’ve done to this giant document makes sense and truly aligns with the audio. Add your formatting as you go, since this is something small and will not take the focus of your attention from double (penta-) checking your work. 

1. Final Adjustment

   1. In a month or so, go back to review the transcript one, final time.

I know, you thought you were done. However, once you’ve taken a step back, started working on another transcript, let your eyes adjust to a different black-and-white document, you will likely come back to make many edits. This is also good since it means you have grown, gotten better, learned enough to improve in just a month. Since you are not so close anymore, it makes it easier to hear some of those unclear words, see that an aside is actually the start of a new sentence, a phrase you never bothered to edit was actually incorrect the whole time (OtterAI is a blessing and a nuisance). This is the smoothest listen-through you will have, for good reason: you put in all that work prior to your sixth, final listen-through.