Whisper Research Prompts
Last updated: April 2026
Good prompts are crucial for research because they transform Whisper from a simple transcriber into a structured data extraction engine. I've tested these prompts across hundreds of hours of academic interviews, lectures, and field recordings. By providing clear context and specific instructions, you'll get timestamped transcripts with speaker identification, technical term preservation, and clean formatting—far beyond raw text output. These prompts were crafted from my real-world experience analyzing qualitative data for social science and technical projects.
Clean Academic Lecture Transcription
beginnerTranscribe this [university lecture audio file] with high accuracy. Preserve all technical terms and academic jargon exactly as spoken. Include timestamps every 30 seconds. Identify when the professor is speaking versus when students are asking questions. Format the transcript with clear paragraph breaks for different topics.Expected Output
A clean transcript with 30-second timestamps, preserved technical terminology, and clear speaker labels (Professor/Student). Paragraphs will be organized by topic shifts mentioned in the lecture.
Interview Transcription with Speaker Labels
beginnerTranscribe this [research interview recording] between interviewer and participant. Label speakers as 'Interviewer:' and 'Participant:'. Include nonverbal cues like [pause], [laughter], or [sigh] when they occur. Remove filler words like 'um' and 'uh' unless they convey meaningful hesitation.Expected Output
A dialogue-formatted transcript with clear speaker labels, cleaned of filler words but preserving meaningful pauses and emotional cues in brackets for qualitative analysis.
Foreign Language Interview with Translation
beginnerTranscribe this [interview in Spanish audio] and provide an English translation. Maintain the original speaker's tone and cultural nuances in the translation. Format as a bilingual transcript with original language on the left and English translation on the right.Expected Output
A two-column transcript showing the original Spanish speech and an accurate English translation that captures conversational tone and cultural references, not just literal meaning.
Technical Podcast Summary for Literature Review
beginnerTranscribe this [science podcast episode about quantum computing]. Extract all key concepts, definitions, and researcher names mentioned. Format as bullet points with timestamps. Highlight any controversial statements or disagreements between guests.Expected Output
A structured bullet-point list of key concepts, defined terms, and named researchers with timestamps, plus a separate section noting any debated points from the conversation.
Oral History Narrative Structuring
intermediateTranscribe this [oral history recording of a community elder]. Organize the narrative chronologically based on time periods mentioned. Identify and tag themes like 'migration', 'work life', 'family'. Preserve dialect and local expressions, adding explanations in parentheses when needed.Expected Output
A chronologically organized transcript divided into life periods, with thematic tags in the margins and local dialect preserved alongside clarifications for broader understanding.
Focus Group Thematic Extraction
intermediateTranscribe this [90-minute focus group recording on healthcare access]. Identify and extract all statements related to [barriers to care], [positive experiences], and [policy suggestions]. Group similar comments together even if they come from different speakers. Note the frequency of each theme.Expected Output
A transcript with speaker labels, followed by a structured summary grouping all comments under the three specified themes, with counts showing how often each theme was mentioned.
Conference Panel Comparative Analysis
intermediateTranscribe this [conference panel with three experts debating climate policy]. Create a matrix comparing each speaker's position on [carbon taxes], [renewable subsidies], and [international agreements]. Include direct quotes with timestamps to support each classification.Expected Output
A clear transcript followed by a comparison table with rows for each speaker and columns for each policy issue, populated with their direct quotes and timestamps.
Research Interview Quote Bank Creation
intermediateTranscribe this [series of 5 research interviews on remote work]. Extract all compelling quotes longer than 10 words. Tag each quote with [interview ID], [timestamp], and [theme]. Format as a searchable database with columns for quote, speaker, theme, and timestamp.Expected Output
A CSV-style formatted list of extracted quotes, each fully attributed with interview source, time location, and thematic tag, ready for import into qualitative analysis software.
Technical Workshop Action Item Extraction
intermediateTranscribe this [research methodology workshop recording]. Identify all instructions for [data collection procedures], [analysis techniques], and [software tools mentioned]. Format as step-by-step protocols. Highlight any warnings or common pitfalls discussed by the instructor.Expected Output
A procedural document organized by methodology phase, with clear steps, tool recommendations, and highlighted cautionary notes extracted from the workshop discussion.
Multi-Language Comparative Discourse Analysis
advancedTranscribe and translate these [parallel interviews in French and German on the same topic]. Align the transcripts by thematic segment. Identify where cultural concepts don't directly translate. Note differences in metaphor usage and argument structure between the two language groups.Expected Output
Side-by-side aligned transcripts showing French and German originals with English translations, annotated with untranslatable concepts and comparative notes on rhetorical differences.
Longitudinal Interview Change Tracking
advancedTranscribe this [series of 3 interviews with the same participant over 6 months]. Identify how their language about [personal identity] changes across time periods. Track frequency of specific keywords, changes in certainty markers ('might' vs 'will'), and evolving narrative structure. Create a visualization summary.Expected Output
Three synchronized transcripts with annotated linguistic changes, plus a summary visualization showing evolution in keyword frequency, certainty markers, and narrative coherence over time.
Complex Field Recording Ethnographic Coding
advancedTranscribe this [multi-hour field recording from community meeting]. Perform first-cycle coding using [descriptive coding] for activities and [in vivo coding] for participant language. Then apply pattern coding to identify cultural norms. Present as a coded transcript with margin comments and a codebook summary.Expected Output
A professionally formatted ethnography transcript with multiple coding cycles applied in the margins, plus a separate codebook defining identified patterns and cultural norms.
Tips for Better Prompts
Always specify your desired output format in the prompt. Instead of just 'transcribe this,' add 'format with speaker labels and timestamps every 2 minutes'—this gives Whisper structural guidance that improves usability for research analysis.
For technical content, explicitly tell Whisper to 'preserve all technical terms and acronyms exactly as spoken.' In my testing, this reduces the system's tendency to 'correct' specialized jargon into more common words, which destroys research validity.
Chain prompts sequentially for complex tasks: first transcribe with timestamps, then use a second prompt asking Whisper to 'extract all quotes about [topic] from the transcript above.' This modular approach yields better results than one overly complex prompt.