Before you record: consent and data handling
Most universities and ethics boards (IRBs, research ethics committees) want participants to know how their recording will be handled. If you plan to use an online transcription service, check that your protocol and consent form allow it. Some committees ask you to name the type of tool or say that audio leaves the device.
What you can tell your committee about RocketScribe: files are never used to train AI models, and you can delete the audio and transcript at any time. Free-plan transcripts are kept for a limited time and paid plans keep them longer. We do not claim HIPAA or GDPR certification, so if your study handles health data or other special categories, ask your data protection office before uploading.
Recording interviews for analysis
Researchers often record with a dedicated recorder that saves WAV, a phone app that saves M4A, or Zoom and Teams for remote participants. All of these can be uploaded directly. Uncompressed WAV files are large; if an upload is slow, the resumable upload keeps going from where it stopped when you press Retry.
For in-person sessions, place the recorder between you and the participant and test the level before you start. For remote sessions, ask the participant to use headphones and record in a quiet room. Note participant IDs at the start of the recording rather than full names. That way the audio file itself carries less identifying information.
Anonymising names before analysis
Most studies need names, places and employers removed or replaced before the transcript goes into coding software or an appendix. Speech recognition will write down every name it hears, so plan an anonymisation pass.
Use the synced player to review the transcript. Click a line to hear it, then edit names in place, for example replacing a person's name with [P04] or a workplace with [Hospital A]. For long transcripts it is often quicker to export DOCX and use Word's find and replace, then keep the anonymised file as your working copy.
If speaker identification is on (paid plans or a Minute Pack, not the Free plan or trial), rename Speaker 1 to "Researcher" and Speaker 2 to the participant code. The new name applies across the whole transcript.
Verbatim or clean transcript?
Methods sections usually state whether transcripts are verbatim (every "um", false start and repetition) or clean (readable sentences with fillers removed). The automatic transcript sits closer to clean: speech recognition tends to drop many filler words and smooth over some false starts.
If your method needs true verbatim, such as conversation analysis or discourse analysis with pauses and overlaps marked, treat the automatic transcript as a first draft. Listen through and add the features your notation requires. For thematic analysis and most interview studies, a clean transcript checked against the audio is usually what supervisors expect.
Exports for coding tools and your thesis
Qualitative analysis tools such as NVivo, ATLAS.ti and MAXQDA accept Word or plain-text documents. Export DOCX or TXT and import it the usual way. There is no direct integration. Export CSV if you want each segment in a spreadsheet row with its timestamp, which some researchers use for their own coding sheet.
Keep timestamps on in your working copy even if you strip them from the version you analyse. When you quote a participant in your findings chapter, the timestamp lets you return to the exact moment in the audio and confirm the wording and tone. Supervisors and examiners sometimes ask how a quote was checked, and this gives you a simple answer.
Use folders to keep one study or one wave of interviews together, and bulk export a folder when you are ready to import. Paid plans allow longer files and batch uploads, which helps when you have a full set of interviews to process. Interviews in other languages work as well, with more than 90 languages supported and mixed Chinese and English handled.