Computer-Generated Transcription – What to know for your next research project
4/7/22 / Maria Cuevas
In this blog post, we explore the pros and cons of using computer-generated transcription features and tools. At Corona, we have experimented with several transcription services and tools, both transcribed by real people and machine. As researchers, we understand that transcripts of focus group conversations and in-depth interviews are essential to every qualitative research project. However, transcript usage can extend beyond research and into our everyday lives as a tool to support facilitating meetings and other important conversations.
Pros of a computer-generated transcription
One of the greatest advantages to using computer-generated transcription, also known as automatic transcription, is convenience. Many computer-generated transcription tools and features are already built into or are compatible with applications that we use every day. Computer-generated transcripts can be available live, in real time, or within hours depending on the length of the audio. This contrasts with manual transcription which is done by people and therefore can take up to a few days. If available, live computer-generated transcription can potentially make a virtual meeting or call more accessible for participants who are hearing impaired or who speak a different language.
Typically, computer-generated transcription has a lower cost than manual transcription. Video calling software and platforms, like Zoom and Teams, have the option to add computer-generated transcription features for an additional cost. There is also data analysis software that includes computer-generated transcription features like NVIVO and Dovetail. These platforms can give users the option to generate verbatim transcripts and include timestamps. Automatic transcripts generated within data analysis software create a more seamless data analysis process and can provide the option to edit any errors in the transcript as you review the materials. Another feature available for computer-generated transcription is the option to transcribe audio in other languages at no additional cost. In contrast, manual transcription service providers often charge more for transcribing in other languages, if they even offer the option.
Cons of a computer-generated transcription
The convenience and lower cost of computer-generated transcription comes with a major tradeoff: it is not as accurate as a manual transcription. Computer-generated transcripts can vary in quality; however, AI test to speech technology is quickly advancing to develop a more accurate transcription process that can detect previously missed nuances. Computer-generated transcripts can be less accurate when speakers have accents, talk fast, have lower audio quality or background noise, or if there are many speakers within the audio. Other inefficacies with automatic transcription occur when multiple languages are spoken within the same audio or one speaker switches between languages. When computer-generated transcripts have many errors, it can become a time-consuming task for the researcher to review and edit them. It is common for computer-generated transcripts to be used as a starting point for manual transcription or simply serve as a guide to an audio or video file.
Which one is right for you?
From a researcher’s perspective, manual transcription is generally worth the cost in order to have a more efficient data analysis process since it is not necessary to spend a great deal of time editing transcripts. Alternatively, using the computer-generated transcription features within data analysis software can allow for a smoother analysis process when doing tasks such as coding data. In general, using computer-generated transcription for a qualitative research project can help keep costs down. Keep in mind that computer-generated transcription is likely better suited for in depth interviews or small groups instead of larger groups where there are many speakers. Computer-generated transcription can be and is widely used beyond research as a tool for notetaking and documenting everyday calls and meetings.
Some of the computer-generated transcription features I have experimented with are on Zoom and data analysis software like Dovetail, NVIVO, and Grain.
Have you tried any computer-generated transcription features or tools? Tell us about your experience in the comments below.