This is an Eval Central archive copy, find the original at evalacademy.com.

Evaluators have several options for transcribing audio from qualitative interviews, including voice-to-text software, outsourcing, and doing it ourselves. Depending on the budget for the project, you may not be able to afford software licenses or the cost to hire someone else to transcribe your interviews, so you might be left transcribing interviews yourself.

If you are taking the DIY approach, here are four tips to get you transcribing like a pro!

Screen Shot 2020-07-09 at 9.31.07 AM.png

1. Decide on the appropriate level of accuracy

Before you start, you need to decide how accurate your transcriptions need to be. You have a range of options for how precise to transcribe, and your choice depends on how the data will be analyzed. Some commonly used levels of accuracy are (from most accurate to least accurate):

Exactly verbatim: Type exactly what is said, including every “um,” “uh,” and “hmmm.”
Intelligent verbatim: Type exactly what is said, EXCEPT for filler words that do not change the meaning. At this level, you would skip phrases like “um,” “you know,” and “like” when appropriate.
Edited transcription: Skip irrelevant sentences that don’t relate to the evaluation. For example, someone might start telling a tangential personal story – if it is not relevant to the purpose of the interview, you might not transcribe it.
Summary transcription: Only important passages are transcribed. For example, you may only require the answers to the questions in the interview guide, and any other comments are skipped.

When choosing a level of accuracy, you face a trade-off between the time it takes and the level of detail. Exactly verbatim transcriptions take the longest to type, while summary transcriptions are the fastest. However, summary transcriptions necessarily exclude some information, so you risk missing important data. It is up to you as the evaluator along with the stakeholders involved to navigate this trade-off between accuracy and resource-use in a way that makes sense for the particular evaluation.

2. Document interview meta-data

At the top of your Word document where you will transcribe the interview, you should include relevant meta-data (i.e., data about the interview). This could be:

Interview ID#
Project name
Participant’s name or initials (remember to maintain appropriate levels of confidentiality)
Interview group, if applicable (e.g., client, program manager, funder)
Interviewer’s name
Date of interview
Location of interview
Transcriber’s name

In the metadata section, also identify any abbreviations that will be used in the transcription. For example, you may simple use “P” to denote the participant, and “I” to denote the interviewer. My transcriptions usually look like this:

I: Can you tell me about your use of the program?

P: Sure, I started using this program when I first moved to the city.

I: When was that?

P: About a year ago.

3. Use audio software

Rather than using the default audio player on your computer, like iTunes or QuickTime, it can greatly speed up your transcribing to use an audio software with more advanced features. I recommend ExpressScribe because it is reasonably priced and has some really important capabilities:

Fine-tuned control over playback speed. It might seem counterintuitive, but slowing down the audio playback speed can actually increase the speed of your transcription. Most of us can’t type as quickly as people talk, so by slowing the audio down to about 50-60%, you will be able to type at the same pace as the speaker. This reduces the amount of time you spend pausing and rewinding the audio!
Global hot-keys. This was a game-changer for me: ExpressScribe allows you to control the audio playback (play, pause, rewind) using keyboard shortcuts that work even when the program is running in the background (that’s why they’re called global hotkeys). The set-up I use is:
- Ctrl/Cmd + down arrow = Stop
- Ctrl/Cmd + up arrow = Play
- Ctrl/Cmd + left arrow = Step back 3 seconds
- Ctrl/Cmd + right arrow = Step forward 3 second

With these advanced controls, you can play/pause and rewind without leaving the Word document you’re typing in!

4. Include time stamps

While you transcribe, it is helpful to include a time stamp in the transcription about every 5 minutes. This allows someone reading the transcript to easily find passages in the audio recording if necessary. I keep an eye on the audio player for when I reach an interval of five minutes, and I document it in the text using square brackets:

I: Can you tell me about your use of the program?

P: Sure, I started using this program when I first moved to the city.

[5:00]

I: When was that?

P: About a year ago.

Time stamps should also be included when you reach a passage you can’t understand. Especially when people refer to place names, people’s names, or words in another language, it can be hard to get the word right. This should be flagged inside square brackets along with the time stamp. For example:

P: After that, we went to [place name 12:05] to gather berries.

After incorporating all of these tips, you can expect verbatim transcriptions to take about 3-6 hours for each hour of recorded audio. Keep this in mind while planning the project: 10 interviews, each an hour long, could take up to 60 hours to fully transcribe!

Email Address

We respect your privacy.

Thank you!