In the last four years of offering software training (MAXQDA, ATLAS.ti, NVivo), one of the top three topics participants ask about is transcription. As such, I have compiled the following questions and answers to guide users’ decisions should they find themselves facing the same doubts.
Q 1: Should I transcribe audio and video files outside of the software, or import them into the software and transcribe within?
A: In general, transcription done outside of research software using a program specifically for transcribing is easier. You can add timestamps when you transcribe either outside or within. One advantage of transcribing within is that the documents will be fully formatted for that software (e.g., MAXQDA, NVivo), but a disadvantage is that the audio or video files take up too much space in your project and may slow it down.
Q2: Do I add timestamps to my transcripts?
A: There is no need to add timestamps to your transcripts if you are never going to use them to pinpoint the exact locations on the audio or video files. If you will only analyze the text of your interviews or focus groups or meetings, you will ensure that the transcriptions are accurate and do NOT have to refer back to the audio or video files using timestamps.
Q3: How should I transcribe interviews or focus groups or conference sessions or webinars or clinical visits, etc.?
A: Each research program has specific requirements on how it will recognize the different data sources when you import them. Be sure to check the manuals of the respective program you are using (MAXQDA, ATLAS.ti, NVivo) before transcribing. This ensures the program can understand what data it is and import it as such.
Q4: Do I transcribe all data sources, or just some of them?
A: This is a rigor question. Due to a limited budget or time, you might choose to selectively transcribe your data for analysis. However, this also introduces biases early on in data analysis because some data is already deemed irrelevant and therefore not transcribed. In cases where you actually code the audio or video files directly in a research program, there may not be any need to transcribe them into text.
Q5: What about the new NVivo auto-transcription feature (released 2019)? Doesn’t it take away the necessity to transcribe manually?
A: This may also be a rigor issue. As of today, this new NVivo feature works best with audio files “with high quality” and the results are “up to 90 percent accurate”. Yes, this feature is significantly cheaper and faster than manual transcription, but it is also far from perfect (like you would expect from human transcriptionists) at the moment. Also, the auto-transcripts will be formatted for use in NVivo only.
Q 1: Should I transcribe audio and video files outside of the software, or import them into the software and transcribe within?
A: In general, transcription done outside of research software using a program specifically for transcribing is easier. You can add timestamps when you transcribe either outside or within. One advantage of transcribing within is that the documents will be fully formatted for that software (e.g., MAXQDA, NVivo), but a disadvantage is that the audio or video files take up too much space in your project and may slow it down.
Q2: Do I add timestamps to my transcripts?
A: There is no need to add timestamps to your transcripts if you are never going to use them to pinpoint the exact locations on the audio or video files. If you will only analyze the text of your interviews or focus groups or meetings, you will ensure that the transcriptions are accurate and do NOT have to refer back to the audio or video files using timestamps.
Q3: How should I transcribe interviews or focus groups or conference sessions or webinars or clinical visits, etc.?
A: Each research program has specific requirements on how it will recognize the different data sources when you import them. Be sure to check the manuals of the respective program you are using (MAXQDA, ATLAS.ti, NVivo) before transcribing. This ensures the program can understand what data it is and import it as such.
Q4: Do I transcribe all data sources, or just some of them?
A: This is a rigor question. Due to a limited budget or time, you might choose to selectively transcribe your data for analysis. However, this also introduces biases early on in data analysis because some data is already deemed irrelevant and therefore not transcribed. In cases where you actually code the audio or video files directly in a research program, there may not be any need to transcribe them into text.
Q5: What about the new NVivo auto-transcription feature (released 2019)? Doesn’t it take away the necessity to transcribe manually?
A: This may also be a rigor issue. As of today, this new NVivo feature works best with audio files “with high quality” and the results are “up to 90 percent accurate”. Yes, this feature is significantly cheaper and faster than manual transcription, but it is also far from perfect (like you would expect from human transcriptionists) at the moment. Also, the auto-transcripts will be formatted for use in NVivo only.