• Home
• About
• Contact
• RSS

Using Audacity to Transcribe Oral History

Not long ago, I finished all the interviews for the-project-that-will-be-my-Masters but was then faced with the daunting task of transcribing them all. Since starting this project I’ve learned about an entire movement within oral history to work from the audio but I’d already committed to sending the texts back out to the participants so they could add or revise. Next time I’ll conceive the project with this in mind. That said, I needed the transcripts in a hurry so I set to fine-tuning my method. This is a run through of my workflow in hopes that it will be helpful to someone else.

First off, I recorded the interviews using my ordinary mp3 player: a Sansa e250 that I’ve had for 3 years or so. It holds 2 GB of data onboard plus it has a microSD card slot if I need it. The voice record option will record up to 4 hours. Since my longest interviews were around 2 1/2 hours I was fine as long as I had time in between to dump the audio to my computer. The battery lasted just fine and the audio quality was good enough with just the onboard mic. All I did was place the mp3 player on a flat surface somewhere in the middle of the people speaking.

Audacity might seem over the top for transcribing software but I chose it for a number of reasons:

• I was already familiar with it from editing music and podcasts
• it’s possible to select small sections of any length and loop the playback
• it has the ability to slow down the playback
• it also has the ability to speed up the playback
• it’s cross-platform (I’d previously used it on Windows but am currently on Ubuntu.)
• and it’s free

I realize there’s transcription software out there and all kinds of fancy stuff like foot pedals and what-have-you but I’m a graduate student. I need to work with what’s available and affordable.

The Sansa recorder makes a .wav file which I imported into Audacity with a simple File > Open. I like to zoom in a bit to make selecting a clip easier. This is done by clicking on the + button.

Next thing to do was use Audacity’s select tool to select a piece of audio by clicking and dragging the mouse across the audio from desired start point to desired end point. Here I’ve selected 30 seconds of audio at the beginning of the recording.

Next I set up my document-that-will-be-the-transcript in OpenOffice with the participant’s name in the header, as well as date and place of the interview.  At the top of the document I list all the people present at the interview. This is what my screen looks like when Audacity is the active window. When I switch back to OpenOffice, Audacity goes behind:

Before getting on with the transcribing the other important keys to know include the Play button for normal playback. You can loop the playback indefinitely if you if you Shift+Click the play button. 

The “p”-key pauses and un-pauses the audio and spacebar stops the playback and returns the cursor to the beginning of the selection. Armed with all these keyboard shortcuts you should be all set. Add Alt+Tab and you can easily switch back and forth between the two programs: listen in Audacity and type in OpenOffice/your favourite wordprocessor.  Once I was in the transcribing-groove, the flow went pretty much like this:

1. Listen in Audacity
2. p to pause
3. Alt+tab back to the text document
4. type what I remember
5. Alt+tab back to Audacity
6. p to un-pause
7. Use the mouse to select the next section of audio
8. Every now and then throw some time stamps into the text for future reference points
9. Lather, rinse, repeat.

When each section was complete I would also listen through to fill in any missing bits or correct any errors.

But wait — there’s more!

Sometimes, especially if there was a lot of laughter (my interviews were fun) or if there were multiple voices speaking at once, it was hard to make out the words. In these cases, it helped to slow down the recording using Audacity’s speed control. Speed control has its own little green playback arrow, located immediately to the left of the slider adjustment. The large green playback arrow in the main control area is only for normal speed playback. “P” for pause and space bar for stop still work. Unfortunately, the looping feature does not.

The most exciting part of the whole process was when I realized I could listen a whole lot faster than the participants and I spoke during the interview. Speeding up the playback meant I could listen to a 60 minute interview in 40 or so minutes. Keyboard shortcuts plus faster playback – even if we sounded like chipmunks – meant I was able to get all the transcribing done in no time at all: specifically just under a week.

And so, that’s it. I’m still looking for more ways to speed up transcribing so if you have any please leave them in the comments. As much as I love oral history and as much I’d like to think I could work directly from audio, truth is I’m visual and I get a lot out of seeing the words on the page.

I would definitely encourage anyone taking on this kind of project to transcribe as you go. I hope I’m never again in a situation where I have to transcribe as much as I did in this short of time. It would have been much easier on me physically (wrists, shoulders, back, etc) if I’d done it in bits and pieces as I’d gone along. I also could have gleaned things from earlier interviews that would have helped the later ones.

Comments