codeck-speech

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

codeck speech

codeck 演讲稿生成

Role activation

角色激活

Read

$DECK_DIR/diagnosis.md

. If a speech role is recommended, use it. Otherwise, pick a coach based on domain and audience:

Technical → Feynman: simplify the complex, bridge with analogy

Business → Jobs: build anticipation, one "one more thing"

Academic → Hans Rosling: let data tell the story

读取

$DECK_DIR/diagnosis.md

。如果推荐使用演讲稿角色，则直接启用。否则，根据领域和受众选择对应风格的导师：

技术类 → Feynman：简化复杂内容，用类比搭建认知桥梁

商业类 → Jobs：营造期待感，设置「one more thing」彩蛋

学术类 → Hans Rosling：让数据讲故事

Setup

环境配置

bash

DECK_DIR="$HOME/.codeck/projects/$(basename "$(pwd)")"
mkdir -p "$DECK_DIR"
bash "$HOME/.claude/skills/codeck/scripts/status.sh" "$DECK_DIR"

Read:

HTML (latest
```
*-r*.html
```
) — actual slide content
outline.md — structure, arc, user intent
design-notes.md — visual intent (speech rhythm should match visual rhythm)

If no HTML and no outline, suggest

/codeck-design

/codeck-outline

first.

If only outline exists, write based on outline — note that the script is based on structure, not final visuals.

Smart skip: skip questions if user's instruction already specifies style and duration.

bash

DECK_DIR="$HOME/.codeck/projects/$(basename "$(pwd)")"
mkdir -p "$DECK_DIR"
bash "$HOME/.claude/skills/codeck/scripts/status.sh" "$DECK_DIR"

读取内容：

HTML（最新的
```
*-r*.html
```
文件）—— 实际的幻灯片内容
outline.md —— 结构、叙事弧、用户意图
design-notes.md —— 视觉设计意图（演讲节奏应匹配视觉节奏）

如果没有HTML也没有大纲，建议先使用

/codeck-design

或

/codeck-outline

命令。

如果只有大纲，则基于大纲撰写——需备注脚本是基于结构而非最终视觉内容生成。

智能跳过： 如果用户指令已经指定了风格和时长，跳过后续询问步骤。

Questions

询问信息

Q1: Style

Q1: 风格

A) TED — conversational, story-driven, breathing room
B) Formal — structured, precise language
C) Casual — natural, humor ok

A) TED —— 口语化、故事驱动、预留停顿空间
B) 正式 —— 结构化、用词精准
C) 轻松随意 —— 自然流畅，可使用幽默

Q2: Duration

Q2: 时长

A) 5 min — lightning, ~1000 words
B) 15 min — standard, ~3000 words
C) 30+ min — deep dive, ~6000 words

A) 5分钟 —— 闪电演讲，约1000字
B) 15分钟 —— 标准演讲，约3000字
C) 30分钟以上 —— 深度分享，约6000字

Generate

生成内容

Before writing, build a fragment map. For each slide in the HTML, list: slide number, title, fragment count (

data-f

elements). This map determines the speech structure — slides with fragments get

### [on enter]

### [fragment N]

sections, slides without get a single block. Do not skip this step.

Write a complete, readable-aloud transcript. Page by page.

撰写前先构建片段映射表。 针对HTML中的每一页幻灯片，列出：幻灯片编号、标题、片段数量（

data-f

元素）。该映射表决定演讲结构——带片段的幻灯片会生成

### [进入时]

### [片段N]

章节，无片段的幻灯片则生成单个内容块。请勿跳过该步骤。

逐页撰写完整、适合大声朗读的逐字稿。

Rules

规则

One section per slide — matches the deck
Transitions — natural bridges between pages
Stage directions — write in the same language as the transcript. Chinese:
```
[停顿 2秒]
```
```
[放慢]
```
```
[看观众]
```
; English:
```
[pause 2s]
```
```
[slow down]
```
```
[look at audience]
```
; other languages: translate accordingly. The speaker must understand them without switching languages.
Word count — ~200 words/min Chinese, ~130 words/min English
Source-based — no fabricated data
Strong opening — story, data, or question
Strong close — callback to opening or call to action

每页幻灯片对应一个章节 —— 与演示文稿结构匹配
过渡衔接 —— 页面之间设置自然的过渡语
舞台提示 —— 与演讲稿使用相同语言编写。中文：
```
[停顿 2秒]
```
```
[放慢语速]
```
```
[看向观众]
```
；英文：
```
[pause 2s]
```
```
[slow down]
```
```
[look at audience]
```
；其他语言对应翻译即可。演讲者无需切换语言即可理解提示内容
字数控制 —— 中文约200字/分钟，英文约130字/分钟
基于原文 —— 不得编造数据
开场有力 —— 用故事、数据或问题开场
收尾有力 —— 呼应开场或给出行动号召

Style notes

风格说明

TED: use "you" / "we", mix short and long sentences, pause after key points, end by echoing the opening.

Formal: complete sentences, logical progression, summarize + outlook at the end.

Casual: colloquial, self-deprecating ok, casual transitions, end with a surprise.

TED风格： 使用「你」/「我们」称谓，长短句结合，关键点后停顿，结尾呼应开场。

正式风格： 句式完整，逻辑递进，结尾总结+展望。

轻松风格： 口语化，可使用自嘲表达，过渡随意，结尾设置惊喜。

Time budget

时间预算

Slide	Title	Words	Estimate
1	...	...	...

A) Help me trim the ones over time
B) I'll manage it myself

幻灯片	标题	字数	预计时长
1	...	...	...

A) 帮我删减超时的内容
B) 我会自行调整

Write back HTML data-notes (fragment-synced)

回写HTML data-notes（与片段同步）

The engine's

buildNotes()

concatenates the slide's

data-notes

with each visible fragment's

data-notes

as the presenter steps through. Use this to sync speech rhythm with fragment rhythm.

引擎的

buildNotes()

方法会在演讲者翻页时，将幻灯片的

data-notes

与每个可见片段的

data-notes

拼接展示。用该功能实现演讲节奏与片段播放节奏的同步。

How it works

工作原理

Read the slide's fragments — find all elements with
```
data-f="N"
```
to know the stepping order
Split the speech into segments — one segment per step (slide entry + each fragment)
Assign notes to each step:
- Slide's
```
data-notes
```
  → what to say when the slide first appears (before any fragment)
- ```
data-f="1"
```
  element's
```
data-notes
```
  → what to say when fragment 1 reveals
- ```
data-f="2"
```
  element's
```
data-notes
```
  → what to say when fragment 2 reveals
- ...and so on

读取幻灯片片段 —— 找到所有带
```
data-f="N"
```
的元素，明确播放顺序
拆分演讲稿为片段 —— 每个步骤对应一个片段（幻灯片进入 + 每个片段展示）
为每个步骤分配备注：
- 幻灯片的
```
data-notes
```
  → 幻灯片首次出现时（展示任何片段前）要说的内容
- ```
data-f="1"
```
  元素的
```
data-notes
```
  → 片段1展示时要说的内容
- ```
data-f="2"
```
  元素的
```
data-notes
```
  → 片段2展示时要说的内容
- ...以此类推

Example

示例

Speech for slide 3:

"Let's talk about the three ideas behind codeck. [pause 2s] First, it recruits people, not rules. [pause] Second, isomorphic mapping. [pause] Third, no schema ceiling."

Slide 3 has

data-f="1"

data-f="2"

data-f="3"

html

<section class="slide" data-notes="Let's talk about the three ideas behind codeck. [pause 2s]">
  <h2 data-f="1" data-notes="First, it recruits people, not rules. [pause]">People, not rules</h2>
  <p data-f="2" data-notes="Second, isomorphic mapping. [pause]">Isomorphic mapping</p>
  <p data-f="3" data-notes="Third, no schema ceiling.">No schema ceiling</p>
</section>

Presenter presses → three times. Notes build up progressively:

Step 0: "Let's talk about the three ideas..."
Step 1: + "First, it recruits people..."
Step 2: + "Second, isomorphic mapping..."
Step 3: + "Third, no schema ceiling."

第3页幻灯片的演讲稿：

"我们来聊聊codeck背后的三个核心理念。[停顿 2s] 第一，以人为本，而非规则约束。[停顿] 第二，同构映射。[停顿] 第三，无schema上限。"

第3页幻灯片有

data-f="1"

、

data-f="2"

、

data-f="3"

：

html

<section class="slide" data-notes="Let's talk about the three ideas behind codeck. [pause 2s]">
  <h2 data-f="1" data-notes="First, it recruits people, not rules. [pause]">People, not rules</h2>
  <p data-f="2" data-notes="Second, isomorphic mapping. [pause]">Isomorphic mapping</p>
  <p data-f="3" data-notes="Third, no schema ceiling.">No schema ceiling</p>
</section>

演讲者按3次翻页键。备注会逐步拼接展示：

步骤0: "我们来聊聊codeck背后的三个核心理念..."
步骤1: + "第一，以人为本，而非规则约束..."
步骤2: + "第二，同构映射..."
步骤3: + "第三，无schema上限。"

Rules

规则

If a slide has no fragments, put the full speech in the slide's
```
data-notes
```
HTML-escape quotes inside
```
data-notes
```
attribute values
Keep stage directions (
```
[pause]
```
,
```
[slow down]
```
, etc.) in the notes
Each segment should be self-contained — the presenter reads what's new at each step
Match the number of speech segments to the number of steps (1 + fragment count)

如果幻灯片没有片段，将完整演讲稿放入幻灯片的
```
data-notes
```
中
```
data-notes
```
属性值内的引号需要进行HTML转义
保留备注中的舞台提示（
```
[pause]
```
、
```
[slow down]
```
等）
每个片段内容独立——演讲者只需阅读每个步骤新增的内容
演讲片段数量与步骤数量（1 + 片段数）匹配

Output: $DECK_DIR/speech.md

输出路径：$DECK_DIR/speech.md

markdown

---
style: "{style}"
duration: "{target}"
totalEstimate: "{estimate}"
---

markdown

---
style: "{style}"
duration: "{target}"
totalEstimate: "{estimate}"
---

Speech: {topic}

演讲稿：{topic}

Slide 1: {title}

幻灯片1：{title}

{verbatim speech text}

[pause 2s]

{逐字演讲内容}

[停顿 2秒]

Slide 2: {title}

幻灯片2：{title}

[on enter]

[进入页面时]

{what to say when slide appears, before any fragment}

{幻灯片加载完成、任何片段展示前要说的内容}

[fragment 1]

[片段1]

{what to say when fragment 1 reveals}

{片段1展示时要说的内容}

[fragment 2]

[片段2]

{what to say when fragment 2 reveals}

{片段2展示时要说的内容}

[fragment 3]

[片段3]

{what to say when fragment 3 reveals}

undefined

{片段3展示时要说的内容}

undefined

Done

完成

Point to the single strongest moment in the script — the line or pause that will land hardest:

codeck speech done.

Strongest moment: {slide N — what happens and why it works. e.g., "Slide 4, the three-second pause after the question. That silence is where the audience decides you're worth listening to."}

{one line — readiness assessment}
Output:
$DECK_DIR/speech.md
+ HTML data-notes updated Press P in the deck for speaker mode to see the script.
All done. Need to export?
/codeck-export
. Check progress anytime with
/codeck
.

指出脚本中最有冲击力的一个瞬间——效果最好的台词或停顿：

codeck 演讲稿生成完成。

最有冲击力的瞬间：{第N页幻灯片 —— 具体内容及效果原因，例如：「第4页幻灯片，提问后的3秒停顿。这段沉默会让观众判断你是否值得倾听。」}

{一行内容——就绪状态评估}
输出：
$DECK_DIR/speech.md
+ HTML data-notes 已更新在演示文稿中按P进入演讲者模式即可查看脚本。
全部完成。需要导出？使用
/codeck-export
命令。随时使用
/codeck
命令查看进度。