B. aTrain

Summary: aTrain is a free, open-source application you run locally on your computer. It will transcribe both your audio and video files, creating text files with speaker names and timestamps. aTrain combines OpenAI’s Whisper transcription models with speaker recognition and provides outputs that integrate with MAXQDA and ATLAS.ti.

Instructions: For Windows, you can download and install it from the Microsoft store. Or for Windows, MacOS, and Linux, you can use the Download aTrain page. Then you just select your audio or video file.Optionally provide information on what language the file is in and how many speakers there are (although speaker identification works better if you specify this). You can also decide what level of model you’d like to use to run the transcription.

aTrain comes installed with the large-v3-turbo model to start, but you can select Models from the left menu and use the download buttons to download other models (generally the larger the model, the more accurate the transcription, but at a sacrifice to speed). You can read more about the models in the documentation.

Once your settings are selected, click on Start. You should see a progress screen. A message will tell you when it is done. You can click on Open to open up a folder with your transcripts created in various formats. If the Open button does not work, you should be able to manually browse to this folder by going to your Documents\aTrain\transcriptions folder. The transcription.txt file just has speaker labels. The transcription_timestamps.txt just has timestamps. The transcription_maxqda.txt has both speaker labels and timestamps.

aTrain also provides a paper with more details on its use.

Technique: Qualitative Data Analysis | Tools: NVivo