Interview - Label speakers
Label the speakers in an unlabeled or messy interview transcript.
Free plan included. No credit card required.
What you provide
2 inputs — previewed below exactly as they appear in Kahubi.
What you get
Kahubi fills this template's expert instruction with your inputs and runs it with a frontier AI model. The result lands as an editable draft in your project — refine it in chat, compile it to PDF, or export it as .docx or .tex. You can also @-mention the template directly in any conversation.
See the instruction behind this template
You are a transcription editor. The transcript below has missing, inconsistent or wrong speaker labels. Reconstruct the turns: (1) infer who is speaking from content cues (questions vs. answers, self-references, names used); (2) output the full transcript with consistent labels on every turn; (3) use the naming hints if given, otherwise label "Interviewer" and "Participant 1/2/…"; (4) keep the wording verbatim — fix only the labels and obvious turn breaks; (5) mark any turn you are unsure about with [speaker uncertain] and list those at the end with your reasoning. Speaker hints (may be "(not provided)"): {{hints}} Transcript: {{transcript}}
Placeholders like {{field}} are replaced with your inputs when the template runs.
Related data templates
Analyze data with framework
Analyze your data through a specific analytical framework you provide.
View templateSuggest survey questions
Generate well-constructed survey questions for your research topic.
View templateCode interview (grounded theory)
Open, axial and selective coding of an interview using grounded theory.
View templateCode interview - you define how
Code an interview transcript using a coding approach you define yourself.
View templateRun “Interview - Label speakers” on your own material
Sign up free — the output is a real draft in your project, not a chat message you have to copy out.