WritingGhostwritingAdvanced45 minSaves 1 hour

Drafting a 20-Minute Keynote: Voice-Matched from Interview Transcripts

Speechwriters supporting executives can draft a 20-minute conference keynote, accurately matching the speaker's unique voice and pacing from raw interview transcripts.

This prompt guides the drafting of a 20-minute keynote speech. It uses interview notes to capture the speaker's specific vocabulary, tone, and rhetorical style, ensuring the final draft aligns with their established public persona. The output includes voice-fidelity notes and alternative openings.

READY-TO-USE PROMPT

Copy Prompt

prompt.txt
As an experienced ghostwriter specializing in executive communications, your task is to craft a keynote speech that authentically reflects the speaker's voice and intent.

### Role
Experienced ghostwriter specializing in executive communications.

### Context
I need a 20-minute keynote speech drafted for `{{speaker_name}}` to be delivered at `{{conference_name}}`. The topic for this keynote is `{{keynote_topic}}`. I have provided a series of `{{interview_notes_text}}` that detail the speaker's thoughts, key messages, preferred vocabulary, and typical communication style. My objective is a speech that feels genuinely from `{{speaker_name}}`, capturing their distinct pacing, choice of words, and overall tone. The content should be informative and authoritative, without resorting to self-promotion.

### Task
1.  **Analyze Speaker's Voice:** Thoroughly review the `{{interview_notes_text}}` to identify `{{speaker_name}}`'s unique speech patterns, recurring phrases, preferred vocabulary, common rhetorical devices, and general tone. Note their typical pacing and how they structure arguments.
2.  **Draft Keynote Speech:** Produce a complete 20-minute keynote speech on `{{keynote_topic}}` for `{{conference_name}}`. The speech must integrate key themes and supporting evidence directly from the `{{interview_notes_text}}`.
3.  **Maintain Tone and Style:** Ensure the draft consistently reflects `{{speaker_name}}`'s identified voice. The speech should be evidence-led, grounding all assertions in data or trends mentioned in the provided notes. The tone must remain ego-suppressing, avoiding any overly promotional or self-congratulatory language.
4.  **Structure:** Organize the speech logically with a clear introduction, well-developed main points, and a concise conclusion.

### Constraints
*   The speech must be precisely 20 minutes in delivery length. (Aim for approximately 2,500-2,800 words, depending on typical speaking pace.)
*   Strict adherence to the voice-matched, evidence-led, and ego-suppressing tone is mandatory.
*   Do not introduce new factual information or data points not explicitly present or clearly implied within the `{{interview_notes_text}}`. If external data would strengthen a point, mark it clearly as a placeholder (e.g., `[INSERT RECENT INDUSTRY STAT HERE]`).
*   Prioritize clarity and conciseness in all phrasing.
*   Incorporate `{{speaker_name}}`'s specific phrasing and common expressions, identified directly from the interview notes, where appropriate.

### Output
1.  **Keynote Speech Draft:** A complete, polished draft of the 20-minute keynote speech.
2.  **Voice-Fidelity Notes:** A separate section detailing 5-8 specific observations on `{{speaker_name}}`'s unique style, vocabulary, and pacing. Explain how these elements were incorporated into the speech and provide specific examples from the drafted text to illustrate each point.
3.  **Alternative Openings:** Two distinct alternative opening paragraphs for the speech. Each opening should offer a slightly different hook while maintaining the speaker's authentic voice and aligning with the overall tone.

Estimated results

DifficultyAdvanced
Setup time45 min
Time saved1 hour
Best modelsChatGPT, Claude, Gemini
Best audienceConsulting, Public Relations

Editor's note

Why this prompt matters

Speechwriters often face the delicate task of capturing an executive's true voice when drafting critical communications. A generic speech, no matter how polished, can undermine a speaker's credibility if it doesn't sound genuinely like them. This workflow is designed for speechwriters who need to translate raw interview notes into a compelling 20-minute keynote that faithfully reflects the speaker's unique communication style, pacing, and vocabulary. It addresses the common challenge of moving beyond mere content synthesis to true voice-matching. This approach is particularly valuable when preparing for high-stakes events like major industry conferences, where an authentic delivery is paramount. Reach for this workflow when you have conducted interviews and need a structured method to produce a nuanced, speaker-centric draft that minimizes extensive revision cycles. It allows you to quickly establish a strong foundational draft that resonates with the executive's personal brand.

Anatomy

Prompt engineering breakdown

Role

As an experienced ghostwriter specializing in executive communications.

Context

Draft a 20-minute keynote for {{speaker_name}} at {{conference_name}} on {{keynote_topic}}, using {{interview_notes_text}} to capture their distinct voice, pacing, and vocabulary. The speech should be informative, authoritative, and avoid self-promotion.

Goal

To produce a complete 20-minute keynote speech that authentically reflects the speaker's voice, integrates key themes from interview notes, and includes voice-fidelity notes and two alternative openings.

Constraints

Speech must be precisely 20 minutes (approx. 2,500-2,800 words). Strict adherence to voice-matched, evidence-led, ego-suppressing tone. Do not introduce new factual information; mark placeholders for external data. Prioritize clarity and conciseness. Incorporate speaker's specific phrasing and common expressions.

Output format

1. Keynote Speech Draft, 2. Voice-Fidelity Notes (5-8 observations with specific examples), 3. Two Alternative Opening paragraphs.

Why this structure works

The prompt utilizes role priming to establish the necessary expertise for the task. Explicit constraints on length, tone, and content ensure the output is precise and meets specific delivery requirements. The structured output format guarantees all required elements, like voice-fidelity notes and alternative openings, are delivered, enhancing the usability for the ghostwriter.

Pick your version

Prompt variations

BeginnerWorks with any model

For users new to voice-matching or when the source material is less complex and a simpler draft is sufficient.

prompt.txt
As a speech assistant, your goal is to help draft a simple 20-minute talk. I need a speech for {{speaker_name}} about {{keynote_topic}} at {{conference_name}}. Use the {{notes_summary}} to get an idea of what they want to say and how they usually talk. Try to make it sound like {{speaker_name}}'s own words. The speech should be easy to follow and focus on the main points. After the speech, list 3-4 key phrases or ways of speaking that you included. Also, give one different way to start the speech. Keep the language straightforward and avoid complex jargon.
ProfessionalBest with chatgpt

When detailed voice-matching and a structured, comprehensive output are critical, mirroring the complexity and requirements of the original prompt.

prompt.txt
Assume the role of a seasoned ghostwriter for executive presentations. Your assignment is to construct a 20-minute keynote address for {{speaker_name}} to be delivered at {{conference_name}}, focusing on {{keynote_topic}}. Review the {{speaker_input_text}} (which includes interview transcripts, previous speeches, and preferred terminology) to deeply understand {{speaker_name}}'s unique cadence, rhetorical style, and vocabulary. The resulting speech must be a true reflection of their voice, maintaining an authoritative yet humble tone. The content should be strictly evidence-based, drawing only from the provided materials. Provide the full speech draft, a separate section detailing 6-7 distinct voice-fidelity observations with textual examples, and two alternative introductory paragraphs that capture a different hook while staying in character. Ensure the speech is approximately 2,500-2,800 words.
Short VersionWorks with any model

For quick drafts or when only the core speech content is needed, without detailed analysis or alternative openings.

prompt.txt
Draft a concise 20-minute keynote speech for {{speaker_name}} on {{keynote_topic}} at {{conference_name}}. Review the {{core_notes}} to accurately capture their distinct voice, including preferred phrases and typical argumentation. The speech should be informative, evidence-led, and avoid any self-promotional language, staying true to {{speaker_name}}'s established tone. Focus on delivering the primary message clearly and directly. Provide the full speech draft in a single output.
EnterpriseBest with claude

In corporate environments where legal, brand, or stakeholder alignment is paramount, requiring compliance and risk assessment.

prompt.txt
Act as a senior ghostwriter specializing in high-stakes corporate communications. Develop a 20-minute keynote speech for {{speaker_name}} regarding {{keynote_topic}} at {{conference_name}}. Critically analyze the {{comprehensive_speaker_profile}} (including interview transcripts, brand guidelines, and previous approved statements) to perfectly replicate the speaker's voice, corporate messaging, and strategic intent. The speech must be rigorously evidence-led, compliant with {{company_brand_guidelines}}, and designed to resonate positively with {{key_stakeholder_groups}}. Identify any potential compliance or reputational risks within the provided content and suggest mitigating language. Output the complete keynote draft, a detailed voice-fidelity and brand compliance report with 7-8 observations, and two alternative openings, each assessed for stakeholder alignment.

What you'll get

Expected output

Keynote Speech Draft Excerpt (Introduction) Good morning, Global Tech Summit. It's a privilege to be here among so many innovators and thought leaders. For years, we've discussed artificial intelligence in terms of its computational prowess, its algorithms, its sheer capacity to process information at scales we once only dreamed of. And rightly so; the technological leaps have been extraordinary. But as we stand at this precipice, contemplating the next wave of AI's integration into our lives, I find myself returning to a fundamental question: Where is the human in all of this? What role do empathy, ethics, and our inherent human biases play in shaping systems that will, undoubtedly, shape us? This isn't a rhetorical question. It's the crux of the matter for responsible AI development. We’ve seen the impressive demonstrations, the efficiency gains, the predictive power. But often, the conversations stop short of truly grappling with the implications for human experience, human dignity, and the very fabric of our societies. My conviction is this: the future of AI isn’t about merely building smarter machines; it’s about building smarter machines *with* smarter human understanding. It's not about if AI will change everything, but how we ensure that change serves humanity’s best interests. Voice-Fidelity Notes 1. Direct Address & Inclusive Language: Dr. Reed often uses 'we' and 'us' to create a shared journey with the audience, as seen in 'we've discussed artificial intelligence' and 'contemplating the next wave.' This fosters a sense of collective exploration. 2. Rhetorical Questions: She frequently poses questions to engage the audience and frame core issues, such as 'Where is the human in all of this?' and 'What role do empathy, ethics... play?'. This prompts reflection rather than simply stating facts. 3. Figurative Language (Metaphors/Analogies): The phrase 'standing at this precipice' is typical of her tendency to use vivid, slightly academic imagery to convey significance. 4. Emphasis on 'Crux' or Core Issue: Dr. Reed often distills complex ideas to their essence, using phrases like 'It's the crux of the matter' to highlight the central point of her argument. 5. Rephrasing for Clarity: Her style includes restating key ideas with slightly different phrasing to ensure understanding and reinforce the message, as in 'It's not about if AI will change everything, but how we ensure that change serves humanity’s best interests.' Alternative Openings 1. Good morning. I want to start by acknowledging something that's often overlooked when we discuss AI: its profound human dimension. We spend so much time marveling at algorithms and processing power, and rightly so. But today, I want to pivot our focus to the designers, the users, and the societies AI impacts. Because the most critical decisions about AI aren't purely technical; they're deeply human. 2. Hello, everyone. Let's be honest: when we talk about artificial intelligence, the conversation can quickly become very technical, very abstract. But my experience has shown me that the true challenge, and indeed the greatest opportunity, lies not in the code itself, but in how we, as humans, choose to wield it. We need to embed our values, our ethics, and our understanding of human behavior directly into AI's DNA. That's the real frontier.

Under the hood

Why this prompt works

This prompt's effectiveness stems from its layered approach to instruction, moving far beyond a simple request for a speech. Role priming establishes the model as an "experienced ghostwriter specializing in executive communications," immediately setting a high standard for output quality and domain-specific understanding. This prevents generic writing and encourages nuanced thinking about executive communication. The inclusion of explicit constraints is critical; specifying delivery length, tone (voice-matched, evidence-led, ego-suppressing), and the strict adherence to provided source material (interview_notes_text) significantly reduces the likelihood of hallucination or off-topic content. Instead of asking for a general speech, the prompt employs task decomposition, breaking down the complex goal into distinct, manageable steps: analyzing voice, drafting, maintaining tone, and structuring. This guides the model through a logical workflow, mirroring a human ghostwriter's process. Finally, requesting a structured output that includes not just the speech draft but also "Voice-Fidelity Notes" and "Alternative Openings" forces the model to not only generate content but also to reflect on its own output, articulate its reasoning, and provide variations. This meta-cognitive element ensures a more refined, analytically sound, and truly useful deliverable compared to a single, undifferentiated text block.

Model fit

Best AI models for this prompt

ChatGPT

ChatGPT models excel at processing and synthesizing large volumes of text, making them suitable for extracting voice patterns and key themes from interview notes. Their ability to generate coherent and contextually relevant prose helps in drafting a structured speech that aligns with the identified style. However, achieving precise voice matching often requires iterative refinement and careful prompt engineering. See the full ChatGPT hub for deeper guidance.

Claude

Claude models are particularly strong in understanding nuance and maintaining contextual integrity, which is crucial for voice-matching. They handle long input contexts well, allowing for thorough analysis of extensive interview transcripts to capture subtle rhetorical flourishes and vocabulary. Users may find Claude effective for initial drafts that require significant stylistic consistency. See the full Claude hub for deeper guidance.

Gemini

Gemini models demonstrate strong multimodal reasoning capabilities, which translate to effective text analysis and generation. Their performance in discerning specific speaking styles and adapting output to match a given persona can be beneficial for this ghostwriting task. Gemini models can produce well-structured speeches while adhering to stylistic constraints, though careful review for subtle stylistic deviations is always advised. See the full Gemini hub for deeper guidance.

When to use

  • When drafting a first pass for a busy executive, synthesizing extensive interview notes into a coherent narrative.
  • For ensuring consistent speaker voice across multiple engagements, especially when working with varied input sources.
  • To quickly generate a foundational draft that requires minimal post-production editing for voice and tone.
  • When the speaker's unique communication style is a critical component of their personal brand and needs to be replicated accurately.
  • When you have rich qualitative data (interviews, previous speeches) that clearly outlines the speaker's preferred vocabulary and pacing.

When not to use

  • If the provided interview notes are sparse, generic, or do not clearly establish a distinct speaker voice.
  • When the primary goal is to generate entirely new, unresearched content rather than synthesizing existing material.
  • For highly technical speeches requiring deep subject matter expertise not explicitly covered or implied in the interview notes.
  • If the speech needs to intentionally deviate from the speaker's known style, perhaps for a specific event or audience.
  • When seeking a speech that heavily features promotional or self-congratulatory language, which this prompt explicitly avoids.

Get more from it

Pro tips

  • 1

    Annotate your source material before inputting to highlight specific speaker quirks, preventing generic output and ensuring precise voice-matching.

  • 2

    Reference a previous speech transcript as a pacing guide for word count adjustments, avoiding an unnatural or rushed delivery.

  • 3

    Iterate on the 'Voice-Fidelity Notes' output; use them to refine subsequent prompts, correcting any subtle voice mismatches.

  • 4

    Provide specific examples of phrases the speaker *would not* use alongside their preferred vocabulary, preventing out-of-character language.

  • 5

    Break down very long interview notes into thematic chunks if context window limits are a concern, preventing data truncation and ensuring full analysis.

Don't ship this

Common mistakes

  • Inputting unstructured or unedited interview notes, leading to a fragmented or inconsistent speech draft.

    Fix — Review and organize notes into key themes and bullet points, highlighting core messages and specific phrasing before input.

  • Expecting perfect factual accuracy or novel data points not present in the original notes.

    Fix — Always treat the output as a draft. Verify all generated facts and fill in `[INSERT...]` placeholders with accurate, current data.

  • Overwriting the prompt's constraints regarding ego-suppressing or evidence-led tone.

    Fix — Trust the prompt's core instructions for tone. Edit post-generation primarily for flow, not to alter the fundamental voice.

  • Not providing sufficient detail about the speaker's typical rhetorical devices or argument structure.

    Fix — Include notes on how the speaker usually introduces points, transitions, or uses anecdotes to enhance structural fidelity.

  • Ignoring the requested 'Voice-Fidelity Notes' output, missing a valuable feedback loop.

    Fix — Review these notes carefully. They offer insight into the model's understanding and can guide refinement for future drafts.

People also ask

Frequently asked questions

Q.How much detail should the `interview_notes_text` contain for optimal results?

The interview_notes_text should be detailed enough to clearly establish the speaker's vocabulary, typical sentence structure, common rhetorical flourishes, and overall tone. Including examples of their unique phrasing and how they connect ideas is beneficial for accurate voice-matching. Aim for depth over sheer volume.

Q.Can this prompt be used for a speech shorter or longer than 20 minutes?

While optimized for 20 minutes (2,500-2,800 words), you can adjust the word count target in the prompt's constraints. Be aware that significant deviations may require more post-generation editing to maintain coherence and pacing. Always specify the new desired length clearly.

Q.What if the speaker has a very informal or highly technical communication style?

The prompt adapts to the style present in your interview_notes_text. If the notes reflect an informal or technical voice, the output will follow suit. The key is to provide representative examples of that specific style for accurate replication.

Q.Will the generated speech be completely original, or does it risk sounding generic?

The speech's originality directly correlates with the specificity and richness of your interview_notes_text. The prompt synthesizes *from your input*, so detailed, unique notes result in a distinct, voice-matched draft. Generic notes will produce a more generalized output.

Q.How should I handle sensitive or confidential information in the interview notes?

Exercise caution with sensitive data. Consider redacting or generalizing highly confidential specifics before inputting the notes. Focus on conveying the speaker's voice and general message, rather than precise, unapproved details. Always verify the final output against security guidelines.

Version 1.0Last reviewed July 8, 2026
Reviewed by PromptInFlow Editorial Team