rhetoric-speech-craft

Speech Draft Writer

One-sentence request → professional speech draft: Based on Aristotle's rhetorical triangle framework (Ethos · Pathos · Logos), generate a speech draft adapted to scenarios such as product launches, annual meetings, and TED-style events, with a built-in pause/tone/pace annotation system and automatic speech duration estimation.

Quick Start

Users only need to specify the speech scenario and topic, and the Agent outputs a complete speech draft with performance annotations following the SOP process:

用户：帮我写一篇年会上的CEO致辞，主题是回顾过去一年的成绩和展望未来
Agent：[按 SOP 流程输出带停顿/语气标注的演讲稿 + 预估时长]

Annotation System Description

Use in the speech draft 【】 to mark speech cues, helping the speaker control rhythm and emotion:

Annotation Category	Example	Description
Pause	`【停顿 2s】`	Pause duration in seconds, used to create suspense or let the audience digest information
Tone	`【语气：激昂】` `【语气：沉稳】` `【语气：温情】`	Switch the emotional tone
Pace	`【语速：放慢】` `【语速：加快】` `【语速：恢复正常】`	Adjust the pace rhythm
Gesture	`【扫视全场】` `【面向左侧观众】` `【走向舞台中央】`	Body language and stage movement cues
Stress	`【重音】`	The next sentence or next word needs additional emphasis
Interaction	`【等待掌声】` `【举手示意观众互动】`	Interaction points with the audience

Duration Estimation Rules

Chinese speech pace varies by scenario (see scenario-specific parameters in Phase 5): product launch 230 characters/minute, annual meeting 220, TED-style 240, ceremonial 200
Average English speech pace:130 words/minute(TED style is about 140-160 words/minute)
Pause time is added separately
Interaction segments (applause, questions) are estimated as additional time based on the annotations
Final output format:预估时长：XX 分 XX 秒（正文 XX 字 + 停顿 XX 秒 + 互动 XX 秒）

SOP Process

Phase 1: Scenario and Requirement Clarification

Goal: Clarify the speech scenario, audience, purpose, and constraints.

Steps：

Identify the speech type, and confirm which of the following scenarios it belongs to:

Type	Typical Scenario	Style Characteristics
Product Launch	Product launch, strategy announcement, brand launch	Confident, forward-looking, tight rhythm, emphasis on Logos + Ethos
Annual Meeting/Celebration	Annual meeting address, anniversary celebration, awards ceremony	Warm, reflective, inspiring, emphasis on Pathos + Ethos
TED-style	Idea sharing, industry insights, personal stories	Story-driven, highly inspiring, emphasis on Pathos + Logos
Ceremonial	Opening/closing remarks, graduation ceremony, inaugural speech	Solemn, concise, ceremonial, emphasis on Ethos

Ask clarifying questions(up to 5 key questions):
- What is the speech occasion and the name of the event?
- What is the speaker's identity/position? (This determines the Ethos strategy)
- Who is the target audience? (employees/customers/investors/the public)
- What is the core purpose of the speech? (inspire/inform/persuade/celebrate)
- What is the target duration? (If no requirement, recommend based on the scenario)
If the user requests to skip clarification, continue based on common default values for the scenario type, and note the assumptions at the beginning of the draft

Output: Speech scenario summary card

📋 演讲场景摘要
- 类型：[发布会/年会/TED式/仪式性]
- 场合：[具体活动名称]
- 演讲者：[身份/职位]
- 听众：[目标受众]
- 核心目的：[激励/告知/说服/庆祝]
- 目标时长：[X 分钟]
- 语言：[中文/英文]

Phase 2: Rhetorical Triangle Analysis (Rhetorical Triangle)

Objective: Based on Aristotle's rhetorical triangle framework, design a persuasion strategy for the speech draft.

Rhetorical Triangle Framework：

         Ethos（可信度）
        /              \
       /    演讲效果     \
      /                  \
  Pathos（情感）——— Logos（逻辑）

Operating Steps：

Ethos Analysis (Building Credibility)：
- What are the speaker's sources of authority? (Position, experience, achievements, professional background)
- How to build trust in the opening? (Self-introduction strategy, shared identity, humble opening)
- At which points throughout the speech does credibility need to be reinforced? (Cite data sources, mention personal experiences)
- Output an Ethos strategy list (3-5 specific methods)
Pathos Analysis (Emotional Resonance)：
- What is the target emotional arc? (Which emotion to start from, what changes to go through, and which emotion to end on)
- What are the core emotional trigger points? (Stories, imagery, contrast, suspense)
- What are the audience's emotional pain points/resonance points?
- Output the emotional arc design:[开场情感] → [转折情感] → [高潮情感] → [结尾情感]
Logos Analysis (Logical Argumentation)：
- What is the core argument? (Summarize the central idea of the speech in one sentence)
- What evidence supports the argument? (Data, cases, analogies, causal reasoning)
- Choice of argumentation structure:
  - Timeline structure (past → present → future)
  - Problem-solution structure (pain point → solution → effect)
  - Tripartite structure (three parallel arguments)
- Output the argumentation framework outline

Output: Rhetorical triangle strategy table

| 维度 | 核心策略 | 具体手段 | 在稿件中的位置 |
|------|----------|----------|----------------|
| Ethos | [策略] | [手段1, 手段2...] | [开场/中段/结尾] |
| Pathos | [策略] | [手段1, 手段2...] | [开场/中段/结尾] |
| Logos | [策略] | [手段1, 手段2...] | [开场/中段/结尾] |

Phase 3: Structural Design and Time Allocation

Objective: Design the overall structure of the speech draft and allocate time for each part.

Operating Steps：

Choose a speech structure model(choose based on the scenario and rhetorical strategy):

Structure Model	Applicable Scenario	Proportion of Each Part
Classic Three-Part Structure	General, annual meeting address	Opening 15% · Body 70% · Closing 15%
Hero's Journey	TED-style, personal story	Ordinary life → Challenge → Low point → Turning point → Transformation → Insight
Problem-Solution-Vision	Product launch, strategy speech	Pain point 20% · Solution 50% · Vision 30%
Monroe's Motivated Sequence	Persuasive speech	Attention→Need→Satisfaction→Visualization→Action
Timeline	Review and summary type	Past 25% · Present 35% · Future 40%

Design the paragraph outline：
- Annotate each paragraph: topic sentence + rhetorical focus (E/P/L) + estimated duration
- Mark the corresponding position of the emotional arc within the structure
- Mark the positions of key pauses and interaction nodes

Time allocation table：

| 段落 | 内容概要 | 修辞重点 | 预估字数 | 预估时长 |
|------|----------|----------|----------|----------|
| 开场 | [概要]   | E/P      | XXX 字   | X 分 X 秒 |
| ...  | ...      | ...      | ...      | ...       |
| 总计 |          |          | XXX 字   | X 分 X 秒 |

Output: Paragraph outline + time allocation table

Phase 4: Speech writing (with annotations)

Goal: Write a complete speech, embedding pause, tone, pace, and gesture annotations.

Operation steps：

Opening writing(Key: the first 30 seconds determine audience attention):
- Choose an opening strategy (based on the scenario):
  - Product launch: suspense style ("Today, we are going to redefine XXX")
  - Annual meeting: gratitude style ("Standing here, looking at every one of your faces...")
  - TED style: story style ("Late one night three years ago, I received an email...")
  - Ceremonial: quotation style ("As XXX said...")
- Opening annotation requirements: must include at least 1 【停顿】 and 1 【语气】 annotation
Body writing：
- Write paragraph by paragraph according to the Phase 3 paragraph outline
- Insert a transition sentence between each paragraph + 【停顿 2-3s】
- Add before key data/arguments 【语速：放慢】 + 【重音】
- Add at emotional climaxes 【语气】 switch annotation
- Use vivid, image-rich language for story paragraphs, annotate 【语速：放慢】
Ending writing(Key: the last sentence will be remembered the longest):
- Choose an ending strategy:
  - Echo the opening (opening and ending correspond)
  - Closing with a memorable line (give one shareable sentence)
  - Call to action (CTA: Call to Action)
  - Vision picture (depict a beautiful future)
- Ending annotation requirements: there must be before the last sentence 【停顿 3s】 + 【语气】
Annotation density control：
- Insert 1 annotation on average every 100-150 characters
- Avoid more than 300 consecutive characters without annotations (the audience will lose the sense of rhythm)
- Annotation density in the opening and ending is higher than in the body

Writing style requirements：

Scenario	Sentence characteristics	Preferred rhetorical devices	Taboos
Product launch	Mainly short sentences, strong rhythm	Parallelism, contrast, rhetorical questions	Lengthy buildup, excessive modesty
Annual meeting	Alternating long and short, warm	Recollection, metaphor, parallelism	Empty slogans, reciting a laundry list
TED style	Conversational, like telling a story	Suspense, imagery, twists	Script-reading tone, piling up jargon
Ceremonial	Solemn and neat, rhythmic	Quotation, antithesis, parallelism	Too colloquial, off-topic

Output: Speech text with complete annotations

Phase 5: Duration estimation and final review

Goal: Accurately estimate the speech duration and conduct a final quality review.

Operation steps：

Duration Calculation：
- Count the total word count of the body text (excluding annotations)
- Select the speech rate baseline based on the speech type:
  - Product launch: 230 words/minute (tight pacing)
  - Annual meeting address: 220 words/minute (composed and steady)
  - TED style: 240 words/minute (natural spoken language)
  - Ceremonial: 200 words/minute (solemn and slow)
- Add up the seconds of all pause annotations
- Add up the estimated time of interactive segments (applause about 5-8s, interactive Q&A about 15-30s)
- Calculate the total duration
Duration Adjustment(if the deviation from the target duration exceeds 10%):
- Over time: mark the paragraphs that can be cut (marked as [可选段落]), and provide streamlining suggestions
- Under time: mark the positions that can be expanded, and suggest adding stories/cases/interactions
Rhetorical Triangle Check：
- Ethos: Is credibility established within the first minute of the opening?
- Pathos: Is the emotional arc complete? Is the climax at the 2/3 point of the whole piece?
- Logos: Is the core argument clear? Is the evidence sufficient?
Annotation Quality Check：
- Does the opening have tone + pause annotations?
- Are there pause annotations at paragraph transitions?
- Are there tone-switch annotations at emotional turning points?
- Are there stress/speech-rate annotations before key data/punchlines?
- Does the ending have pause + tone annotations?

Output：

📊 时长估算报告
- 正文字数：XXX 字
- 语速基准：XXX 字/分钟
- 正文朗读时长：X 分 X 秒
- 停顿累计：X 秒（共 X 处）
- 互动预留：X 秒（共 X 处）
- 预估总时长：X 分 X 秒
- 目标时长：X 分钟
- 偏差：[±X%，符合要求 / 需要调整]

✅ 修辞三角检查
- Ethos（可信度）：[通过/需优化] — [简要说明]
- Pathos（情感）：[通过/需优化] — [简要说明]
- Logos（逻辑）：[通过/需优化] — [简要说明]

✅ 标注完整性检查
- 标注总数：X 个
- 标注密度：约每 XXX 字 1 个标注
- 开场标注：[完整/缺失]
- 过渡标注：[完整/缺失]
- 结尾标注：[完整/缺失]

Process Control Rules

Interaction Mode Selection

User Input	Mode	Behavior
Only scenario and topic (< 50 words)	Guided mode	Execute Phase 1 questioning, wait for the user's answer before continuing
Some details (scenario + audience + purpose)	Semi-automatic mode	Ask 2-3 key questions while starting rhetorical analysis
Detailed brief (> 200 words)	Fully automatic mode	Go directly to Phase 2, skip clarification
User says "just write it/don't ask"	Quick mode	Output the complete speech directly based on reasonable assumptions

Iterative Optimization

If the user gives feedback on the speech:

Change content: Return to Phase 4 to modify the corresponding paragraphs, and cascade-update the duration estimate
Change structure: Return to Phase 3 to adjust the structure, and rewrite the affected paragraphs
Change strategy: Return to Phase 2 to adjust the rhetorical triangle strategy, and re-run Phases 3-5
Change scenario: Return to Phase 1 to re-clarify, and re-run the entire process

After all modifications, the duration estimation and quality check of Phase 5 must be re-executed.

Reference Methodologies

Aristotle's Rhetorical Triangle(Aristotle's rhetorical triangle): A three-dimensional persuasion framework of Ethos (credibility), Pathos (emotion), and Logos (logic)
Monroe's Motivated Sequence(Monroe's motivated sequence): Attention → Need → Satisfaction → Visualization → Action
Hero's Journey(Hero's journey narrative structure): Joseph Campbell's story arc, suitable for TED-style speeches
Rule of Three(Rule of Three): In a speech, three parallel points are the easiest to remember
Peak-End Rule(Peak-End Rule): Listeners mainly remember the emotional peak and the ending, which guides the allocation of annotation density