Generate engaging, viral-potential video titles from SRT subtitle files.
Description
A skill that analyzes SRT subtitle/transcript files and generates compelling video titles optimized for YouTube, Xiaohongshu (小红书), and other platforms. Uses AI to identify key hooks, emotional triggers, and viral elements from the content.
When to Use
Use this skill when users:
- Provide an SRT file and want to generate video titles
- Mention "生成标题", "创建标题", "视频标题", "title from transcript"
- Want to create YouTube/小红书/Bilibili titles from transcripts
- Need multiple title variations for A/B testing
- Want platform-specific title optimization
Features
- Multi-Platform Support: YouTube, 小红书, Bilibili, 抖音 optimized titles
- AI-Powered Analysis: Deep content analysis for hook identification
- Viral Optimization: Emotional triggers, power words, curiosity gaps
- Character Limits: Respects platform-specific length requirements
- Bilingual: Chinese and English title generation
- Multiple Variations: Generate 3-5 title options per request
Usage
Basic Workflow
- Read the SRT file to get the transcript content
- Apply the analysis prompt to identify key elements
- Generate titles based on platform requirements
Platform Specifications
| Platform | Max Length | Style |
|----------|------------|-------|
| YouTube | 60 chars | Hook + Keywords, 数字/问句/大胆声明 |
| 小红书 | 20 chars | 简短爆点 + emoji |
| Bilibili | 80 chars | 详细 + B站风格 |
| 抖音 | 55 chars | 悬念 + 口语化 |
Title Generation Prompt
When generating titles, use this analysis framework:
<scratchpad>
- 主题识别: 从字幕中提取核心主题和关键点
- 钩子发现: 找出最吸引人的角度或故事
- 点击欲望: 什么会让观众想要点击
- 草拟标题: 列出 3-5 个候选标题
- 最优选择: 根据以下标准选出最佳:
- 吸引眼球、制造好奇
- 简洁 (YouTube ≤60字符)
- 包含相关关键词
- 准确反映内容
- 使用情绪触发词
- 避免标题党
</scratchpad>
Title Formulas (标题公式)
Hook Patterns (钩子模式)
- 数字法: "3个方法让你..."、"99%的人不知道的..."
- 问句法: "为什么...？"、"如何...？"
- 反差法: "我曾以为...直到..."
- 时效法: "2024最新..."、"刚刚发布..."
- 悬念法: "结果让我震惊..."、"最后一个太绝了"
- 共情法: "打工人必看"、"学生党福音"
- 权威法: "专家揭秘"、"官方确认"
Power Words (力量词汇)
中文:
- 紧迫: 必看、速看、刚刚、最新、限时
- 好奇: 真相、揭秘、原来、竟然、居然
- 价值: 干货、攻略、避坑、省钱、效率
- 情绪: 炸裂、绝了、太强了、神级、王炸
English:
- Urgency: NOW, FINALLY, BREAKING, NEW
- Curiosity: SECRET, TRUTH, REVEALED, HIDDEN
- Value: FREE, EASY, ULTIMATE, COMPLETE
- Emotion: INSANE, MIND-BLOWING, GAME-CHANGER
Output Format
Generate output in this structure:
## 分析过程
核心主题
[从字幕提取的1-2句主题总结]
关键钩子
- 钩子1: [描述]
- 钩子2: [描述]
目标受众
[谁会对这个内容感兴趣]
---
推荐标题
YouTube (≤60字符)
- 首选: [标题]
- 备选: [标题]
- 备选: [标题]
小红书 (≤20字符)
- [标题 + emoji]
- [标题 + emoji]
Bilibili (≤80字符)
- [标题]
- [标题]
---
标题解析
[解释为什么首选标题最有效]
Example
Input SRT Content (excerpt):
本期节目我们将进行一个Vibe Coding的极限挑战
你将看到我们如何用两句大白话就让Claude徒手写出一个下载播客的Skills
Generated Titles:
YouTube:
- 首选: 两句话让AI写出完整程序！Vibe Coding极限挑战
- 备选: 不会编程也能开发？我用Claude徒手造了个下载器
- 备选: AI编程有多强？2分钟写完专业程序员3天的活
小红书:
- 两句话让AI写程序🤯
- 不会代码也能开发✨
Bilibili:
- 【实测】只用大白话就能让Claude写出完整程序？Vibe Coding极限挑战全记录
- 零基础编程时代来了！手把手教你用AI造自己的工具
Claude Integration
When user provides an SRT file for title generation:
- Read the SRT file:
cat "path/to/file.srt"
- Extract text content (remove timestamps):
- Strip SRT formatting (numbers, timestamps)
- Concatenate dialogue/narration text
- Analyze and generate using the prompt framework above
- Present titles in the structured format
Tips for Best Results
- Provide context: Tell Claude the target platform
- Specify tone: 专业/轻松/搞笑/悬疑
- Include keywords: 如果有特定SEO关键词
- Mention competitors: 参考哪些成功标题
Limitations
- Title quality depends on transcript content richness
- Cultural/platform-specific nuances may need human review
- SEO keyword research should supplement AI suggestions
- A/B testing recommended for final selection
Version History
v1.0 (Current)
- Initial release
- Multi-platform title generation
- Chinese and English support
- Analysis framework and power words library
