Focus group transcription
Focus groups are the hardest recordings to transcribe by hand: many voices, interruptions and sentences someone else finishes. Themera separates speakers automatically and gives you an editor to put every passage under the right name.
Transcribe 30 minutes free. No card needed.
Why groups are harder than one-to-one interviews
Speaker separation tells voices apart but does not know who people are. Similar voices can be merged, and one person can be split into two speakers. When several people talk at once, the model provider notes that usually only one voice is transcribed.
So expect more checking than for a single interview, especially in lively parts of the discussion. Timestamps take you straight to those passages.
Recording a group so it transcribes well
- Use one good microphone in the middle of the table rather than a phone at the edge, and test it with two people first.
- Pick a quiet room without echo, music or traffic noise.
- Start with a round in which everyone says their code or pseudonym. It makes matching voices much easier later.
- Ask people to speak one at a time, and as moderator briefly restate passages that were hard to follow.
Assigning and fixing speakers
In the editor, rename "Speaker 1", "Speaker 2" and so on, for example to M for the moderator and P1 to P6 for participants. Give two detected speakers the same name and they are merged. For a single passage, choose the speaker right there, and click its timestamp to hear it.
Mark the moderator as interviewer if you analyse in Themera: their questions are then left out of the coding.
Analysis: the group or the individual?
Whether you compare the group as a whole or individual participants depends on your question. For participant-level analysis, export the transcript as CSV with a speaker column. The focus group analysis guide covers the method.
Frequently asked questions
How many speakers can it detect?
We do not promise a fixed number. With similar voices, always check the assignment; merging and renaming takes a few clicks in the editor.
What happens with crosstalk?
Usually only one voice is transcribed. Listen to those passages via their timestamps and add the missing words by hand.
Can the moderator be left out of the analysis?
Yes. Mark the moderator as interviewer; their passages stay in the transcript but are not coded.
Related
Further reading
Transcribe 30 minutes free
No card needed. Upload a recording, check the transcript, download it as Word, text or for MAXQDA.