The most common feeling when faced with all the intricacies of making a podcast accessible is overwhelm, often leading to shutting down or an inescapable guilt spiral as your podcast continues to not have transcripts. Production workflow is a tricky thing, especially if you’re trying to back-fill the part that should have been the accessibility step. Take Radio Drama Revival for instance: we’re in our fourteenth year, my third year as being part of the team, and we’ve only just figured out how to insert a transcriptionist into our team so that transcripts for all our episodes can come out on time. (Shoutout to Katie Youmans!). And of course, accessibility doesn’t just stop at transcripts, as recently discussed in Caroline Minck’s article at Sound Profitable on making podcasts more accessible.
Transcripts: Planning Ahead & Course Correcting
There are several resources on how to format your transcripts, especially for fiction podcasts. Cassie Josephs at Discover Pods, for instance, goes into detail on this subject, underlining the fact that your script for your actors does not function as a transcript and guides readers through every step necessary for transforming a recording script into a transcript. Granted, when you’re working with an improvised fiction show--actual play, for instance, or improvised fiction like Midst--the steps are more complicated, as they often involve creating a transcript from scratch.
So, if you know how to format the transcript and you have a vague idea of how long it’ll take, how do you adapt the workflow to fit?
First, take a look at this slide from Erin Kyan and Lee Davis-Thalbourne’s talk at Audiocraft 2018, where they discussed the production workflow for Love and Luck, which is produced a season at a time.
They’ve planned time for accessibility needs from the beginning in order to have a realistic timeline for production where they aren’t skimping on the captions that they offer on YouTube for their show. This is a much simpler thing to handle if you’re still in the planning stages, or if you produce your podcast entirely before launching, in a front-loading style. Consider what kind of accessible work you’re going to offer (transcripts and/or captions) and add time before your launch date.
Put a pin in this. Let’s talk about what happens if you are trying to course-correct, or if you produce your podcast in a more cyclical fashion, like an episode or half a season at a time.
Every podcast has to have some kind of production workflow and cycle. One of the several versions we’ve gone through for Radio Drama Revival’s episodes looks like this:
This is our cyclic production schedule for every pair of episodes; the time-length varies due to scheduling for interviews. When considering how transcripts would work, we knew we’d want to give our transcriptionist the most amount of time possible for dealing with interviews, and that she could do interview transcription before the scripted part of the show was completed. Our transcriptionist has, at the outset, two passes at an interview episode: once when the interview edits are complete, and once when the final mix is complete.
However, we also run a Patreon where we offer extended interview cuts, which means our transcriptionist also does a third pass: she transcribes the Patreon extended cut first and then goes through it with the public feed audio to cut whatever didn’t make it out of the Patreon feed. Internally, we’ve set deadlines for everything with our transcriptionist’s timeline in mind so that she has the maximum amount of time available to her to get it done. (i.e., “Interview audio must be handed in by this date, so that the editor has this long to work on it and submit the edited version by this date, so that the transcriber can start working.”).
And now we come to the point about “time”; retrieve the pin from earlier. There are some general guidelines you can find on timing; Mincks smartly recommends planning 2-3x the length of the audio for fixing an automatically generated transcript. But it will take people wildly different amounts of time depending on factors like:
- how many people are in the audio,
- what kind of accents they’re talking in (most automated services hate anything that isn’t a native-sounding West Coast American accent),
- how much overlap there is in dialogue,
- the quality of the audio,
- whether the person has transcribed before (finding a transcription flow is key, and if this is your first time, it will take you longer),
- what kind of other disabilities you’re working with (I have ADHD and transcription is one of the tasks that causes me physical pain because it is so inextricably boring),
- and still more.
If you hire a transcriber to join your team, they’ll likely have an idea of how long transcription will take them already. If, however, you or your transcriber haven’t done transcription before, my best recommendation is to test an episode. Set aside a day or two where you take either one of your show’s episodes if you’ve already produced some, or episodes from a show similar to yours in length and structure if you’re in the planning stages, and test it. Set a timer and start transcribing the way that you would for your show, either from scratch or from an automatically generated transcript. How long does it take you to achieve the finished state?
And once you have that, you can determine: How feasible is it to do this cyclically, and when would you need completed audio every cycle in order to achieve that? How many episodes do you have and how long will it take to complete transcripts for all of them? When is the best point in your production cycle to start transcription?
When you’re trying to course correct, I recommend that you first put a plan in place for having transcripts from here on out. Edit the workflow with your team, change your deadlines to earlier, and find what needs to happen when in order for transcripts to go live at the same time as the audio for the rest of the future of your show. It will likely take a couple of tries, but once that flow is in place and you’ve completed some transcripts with minimal hitches, then bring on the backlog. Talk with your transcriber, if it isn’t yourself, and check schedules to see what’s a feasible timeline for tackling old episodes on top of the future ones: one a week? two a month? Remember the timing from the test episode; it’ll be extremely useful here.
But of course, transcripts are not the be-all, end-all of accessibility.