How to transcribe a research interview
Decide how much of the conversation to preserve before you edit. Then check the transcript against the recording and use consistent speaker labels.
Choose an editing convention first
A readable interview and a detailed record of speech are different outputs. For a readable version, you might remove repeated filler words while preserving meaning. For close analysis of speech, hesitations, interruptions and repetitions may matter. Agree on that convention with the person supervising or using the research.
Automatic transcription does not replace that decision. Do not call a cleaned document a verbatim record, and do not assume an automatic transcript has captured every pause or non-speech sound.
Prepare the recording
- Keep the original file and a note of the interview date.
- Check the recording permissions and the requirements of your project before uploading participant material to any service.
- Choose the language spoken in the interview. Transcription does not translate it.
- Decide how long the uploaded audio should remain available. TranscribeCat offers a retention choice at checkout; its privacy page explains the details.
Label speakers without guessing
An automatic label such as “Speaker A” is a proposed voice grouping, not a person’s identity. Listen to the beginning of each speaker’s contribution before renaming it. You can use “Interviewer” and “Participant 01” if names do not belong in the working document.
If two voices are grouped together, fix the affected passages. Overlapping speech and similar voices can confuse automatic labels. Check that a label still belongs to the same person near the end of the recording.
Review the passages you will analyze
Listen while reading. Correct names and specialist terms, then revisit any sentence whose meaning seems uncertain. If you cannot hear a passage, record that uncertainty and its timestamp rather than writing what you think the participant probably meant.
Before using an exact quotation, check the wording and speaker against the audio again. Keep a note of any edits to the quotation. The review checklist gives a repeatable three-pass process.
Export a working document
Save the edits before downloading. DOCX, TXT and PDF include saved text changes; PDF is available for supported writing systems. SRT and VTT provide timestamps and speaker renames but use the original wording, so correct their text separately before importing it. Check the requirements of your analysis software; a downloadable file is not a dedicated integration.
For a one-hour interview, TranscribeCat costs $2.00. The $2.00 minimum applies to each file, so six ten-minute interviews cost $12.00. See pricing before deciding whether per-file transcription fits your project, and inspect the real sample output before uploading participant recordings.