video-podcast-maker

Video Podcast Maker

中文文档

Automated pipeline to create professional video podcasts from a topic. Supports Bilibili, YouTube, Xiaohongshu, Douyin, and WeChat Channels with multi-language output (zh-CN, en-US). Combines research, script generation, local TTS (edge free, plus azure), Remotion rendering, and FFmpeg mixing. Current release: v5.3.0 — see CHANGELOG.md for version history.

Works with: Claude Code · OpenClaw · OpenCode · Codex · Pi — any coding agent that supports SKILL.md

Publish to: Bilibili · YouTube · Xiaohongshu · Douyin · WeChat Channels

No coding required! Just describe your topic in plain language — the coding agent guides you through each step interactively. You make creative decisions, the agent handles all the technical details.

Note: This project is still under active development and may not be fully mature yet. Your feedback is greatly appreciated — feel free to open an issue.

Features

• Topic → 4K video - research, narration script, TTS audio, Remotion composition, 4K render + BGM in one pipeline
• Local TTS backends - Edge (free, no key) and Azure — synthesized in-house, no external component skill required
• Asset engine - per-video manifest with license provenance; producers are user files, assetseeker stock, imagencn AI stills, videogencn AI B-roll, and Hyperframes overlays — paid generation always asks first
• 4K output + Remotion-native subtitles - 3840×2160; SRT rendered in React at 4K (legacy FFmpeg burn-in available)
• Design learning - extract style profiles from reference videos/images; auto-applied when topics match
• Vertical shorts - 9:16 highlight clips generated from long-form sections
• Multi-platform & multi-language - Bilibili / YouTube / Xiaohongshu / Douyin / WeChat Channels × zh-CN / en-US, with per-platform publish info
• Pronunciation control - global + per-project phoneme dictionaries for Chinese polyphones

Quick Start

1. Install: with the skills CLI, pointing at the full skill:

npx skills add Agents365-ai/video-podcast-maker/skills/video-podcast-maker -g

Drop the /skills/video-podcast-maker suffix to install all three variants (full, -lite, -nano), or clone this repo instead. Paths below are written from the repo root; under a skills CLI install the same files live in the agent's ${SKILL_DIR}.

2. Set up — Python 3.8+, Node.js 18+, FFmpeg, and a Remotion project:

brew install ffmpeg node python3          # macOS (Ubuntu: sudo apt install ffmpeg nodejs python3)
pip install -r skills/video-podcast-maker/requirements.txt
npx create-video@latest my-video-project   # or reuse an existing Remotion project
cd my-video-project && npm i

One-time cost: a fresh Remotion project downloads ~2.2 GB of npm packages plus a ~90 MB Chrome headless shell. Prefer reusing an existing Remotion project (with node_modules/ already installed) for your next video — the heavy install happens once per project, not per video. Lottie animations are optional (@remotion/lottie + lottie-web); install them per project only if you use LottieAnimation.

3. Configure — set TTS_BACKEND plus its API keys (see TTS Backends and Environment Variables).

4. Tell your agent:

"Create a video podcast about [your topic]"

The agent runs the whole workflow (research → script → TTS → Remotion composition → Studio review → 4K render + BGM). Preview and iterate in Remotion Studio (npx remotion studio src/remotion/index.ts); the agent waits for your explicit "render 4K" confirmation before the final render.

⚠️ For the human reading this (not the AI): manually polish podcast.txt, repeatedly

This section is for you, the human — not the agent. Every downstream step — TTS narration, subtitles, section transitions, animation timing, final cut — is derived from this single podcast.txt. A weak script renders into 4K garbage. No amount of polish downstream saves it.

The AI-generated draft is a starting point, nothing more. Do these yourself — don't hand them off to the AI:

1. Mentally read it as the narrator. Treat each sentence as one breath — if a line forces you to "catch your breath" or backtrack to parse, fix it. Where you stumble silently is where TTS stumbles audibly.
2. Revise at least three times.
   • Pass 1: typos, awkward phrasing, tongue-twisters
   • Pass 2: cut filler, cut throat-clearing intros ("So today we're going to talk about…"), cut redundancy
   • Pass 3: tune rhythm — where to pause, where to break a long sentence, which word carries the stress
3. Read each [SECTION:xxx] block end-to-end. Confirm each section opens with a hook and lands a clean transition into the next — not a bullet-point dump.
4. Audit numbers, proper nouns, and English terms separately. ~90% of TTS mispronunciations live here. If pronunciation is wrong, add it to phonemes.json; if it just sounds awkward, rewrite it.
5. Know your length budget. Estimate ~280 zh-CN chars/min or ~150 en words/min. A 5–10 min video means ~1400–2800 chars / 750–1500 words. Don't pad to fill time.

The only acceptance test: read through it once in your head — does any line make you wince? If yes, don't move on to Step 7 (TTS) yet. Otherwise you're just rendering 4K of something even you don't want to hear.

Workflow

Related Skills

Variants in this repo (skills/):

• video-podcast-maker — the full production pipeline (this README's subject)
• video-podcast-maker-lite — minimal personal pipeline: Azure SSML TTS + Remotion, no bundled templates
• video-podcast-maker-nano — tool-agnostic, logic-only pipeline (any TTS backend, any video tool); autonomous by default, oversight configured per project

External skills:

• remotion-best-practices - recommended; core Remotion patterns and guidelines (built-in minimum rules if absent)
• assetseeker - optional; license-vetted stock photos/video/BGM/SFX/icons/fonts
• imagencn - optional; AI stills and thumbnails (paid APIs)
• videogencn - optional; AI video clips for B-roll (paid APIs)
• Hyperframes - optional; transparent overlay animations (Node 22+)

Requirements

Software	Version	Purpose
macOS / Linux	-	Tested on macOS, Linux compatible
Python	3.8+	TTS script, automation
Node.js	18+	Remotion video rendering
FFmpeg	4.0+	Audio/video processing

Installed through the skills CLI? SKILL.md, scripts, and templates then live under the agent's ${SKILL_DIR}; paths in this README are written from the repo-root perspective, which is what a clone gives you.

TTS Backends (local)

TTS synthesis is in-house — no external component skill required. Set TTS_BACKEND to a platform id; only the active platform's env vars are needed:

TTS_BACKEND	Provider	Required env vars	Get Key
edge (default)	Microsoft Edge TTS	(none — free)	—
azure	Microsoft Azure Speech	AZURE_SPEECH_KEY, AZURE_SPEECH_REGION (default eastasia)	Azure Portal

Want more platforms? The former ttscn component skill (cosyvoice, doubao, tencent, baidu, minimax, xunfei, elevenlabs, openai, google) is no longer a dependency. Install it separately and call it directly if you need those.

Environment Variables

Add to ~/.zshrc or ~/.bashrc:

export TTS_BACKEND="edge"                  # edge (default) / azure
export TTS_VOICE="zh-CN-XiaoxiaoNeural"    # optional; unset = backend default
export TTS_RATE="+5%"                      # optional; also settable in user_prefs.json (global.tts.rate)
export TTS_STYLE="gentle"                  # optional; azure only
export AZURE_SPEECH_KEY="..."              # keys for azure (see table above)
export AZURE_SPEECH_REGION="eastasia"      # azure speech region
export GEMINI_API_KEY="..."                # optional: AI thumbnails (imagencn)
export DASHSCOPE_API_KEY="..."             # optional: AI thumbnails (imagencn; ark/hunyuan/zhipu/step also work)

Then reload: source ~/.zshrc

Configuration

Mutable user-level files live in ~/.video-podcast-maker/ (shared across projects, safe from skill updates); the rest live in the skill root (skills/video-podcast-maker/ in this repo, ${SKILL_DIR} when installed):

File	Location	Purpose
phonemes.json	~/.video-podcast-maker/	Global polyphone dictionary; auto-created from the bundled template; per-project overrides in videos/{name}/phonemes.json
user_prefs.json	~/.video-podcast-maker/	Your preferences (TTS, BGM, platform, visual overrides, style profiles); auto-created from template
user_prefs.template.json / phonemes.template.json	Skill root	Default templates — sources for the user-level copies
prefs_schema.json	Skill root	JSON Schema for preference validation
tsconfig.json	Skill root	TypeScript config for Remotion templates

Output structure — every video renders into its own videos/{name}/ directory:

videos/{video-name}/
├── topic_definition.md      # Topic direction
├── topic_research.md        # Research notes
├── podcast.txt              # Narration script
├── phonemes.json            # (Optional) pronunciation overrides
├── assets/manifest.json     # Asset registry (role / source / license)
├── podcast_audio.wav        # TTS audio
├── podcast_audio.srt        # Subtitles
├── timing.json              # Section timing (drives animation sync)
├── thumbnail_*.png          # Video thumbnails
├── publish_info.md          # Title, tags, description
├── output.mp4               # Raw 4K render
├── video_with_bgm.mp4       # With BGM
├── bgm.mp3                  # Background music
├── final_video.mp4          # Final output
└── shorts/                  # (Optional) 9:16 vertical shorts

Background music: bundled tracks live in skills/video-podcast-maker/assets/ — perfect-beauty-191271.mp3 (upbeat) and snow-stevekaldes-piano-397491.mp3 (calm piano). Per-platform behavior (thumbnails, chapters, CTA, publish formats) is documented in the skill's references/platform-matrix.md.

❤️ Support

If this project helps you, consider supporting the author:

WeChat Pay

Alipay

Buy Me a Coffee

Give a Reward

👤 Author

Agents365-ai

• Bilibili: https://space.bilibili.com/441831884
• GitHub: https://github.com/Agents365-ai

📄 License

MIT — Permission is hereby granted, free of charge, to any person obtaining a copy of this software.

视频播客生成器

English

自动化流程，从主题生成专业视频播客。支持 B站 (Bilibili)、YouTube、小红书、抖音和微信视频号，多语言输出（zh-CN、en-US）。集成研究、脚本撰写、内置本地 TTS（edge 免费，外加 azure）、Remotion 渲染和 FFmpeg 混音。当前版本：v5.3.0 —— 版本历史见 CHANGELOG.md。

支持工具： Claude Code · OpenClaw · OpenCode · Codex · Pi — 任何支持 SKILL.md 的 coding agent

发布平台： B站 · YouTube · 小红书 · 抖音 · 微信视频号

无需编程！ 用自然语言描述你的主题，coding agent 会一步步引导你完成。你做创意决策，agent 处理所有技术细节。

提示： 本项目仍在持续迭代完善中，部分功能可能还不太成熟。欢迎提出宝贵意见和建议 — 可以 提交 Issue 或直接联系作者！

功能特点

• 主题 → 4K 成片 - 研究、旁白脚本、TTS 音频、Remotion 合成、4K 渲染 + BGM 一条龙
• 本地 TTS 后端 - Edge（免费，无密钥）和 Azure —— 内置合成，无需外部组件技能
• 资产引擎 - 每视频 manifest 记录角色/来源/许可；生产者包括用户文件、assetseeker 图库、imagencn AI 图片、videogencn AI 视频、Hyperframes 透明叠层——付费生成必先确认
• 4K 输出 + Remotion 原生字幕 - 3840×2160；SRT 在 Remotion 内以 React 4K 渲染（传统 FFmpeg 烧录仍可用）
• 设计学习 - 从参考视频/图片提取风格档案；主题匹配时自动套用
• 竖屏短片 - 从长视频章节生成 9:16 精华片段
• 多平台 & 多语言 - B站 / YouTube / 小红书 / 抖音 / 微信视频号 × zh-CN / en-US，按平台生成发布信息
• 发音控制 - 全局 + 项目级多音字词典

快速开始

1. 安装： 用 skills CLI 安装完整版技能：

npx skills add Agents365-ai/video-podcast-maker/skills/video-podcast-maker -g

去掉 /skills/video-podcast-maker 后缀会安装全部三个变体（完整版、-lite、-nano）；也可以直接克隆本仓库。下文路径均按仓库根目录书写，通过 CLI 安装时同一批文件位于 agent 的 ${SKILL_DIR} 下。

2. 环境准备 — Python 3.8+、Node.js 18+、FFmpeg，以及一个 Remotion 项目：

brew install ffmpeg node python3          # macOS（Ubuntu: sudo apt install ffmpeg nodejs python3）
pip install -r skills/video-podcast-maker/requirements.txt
npx create-video@latest my-video-project   # 或复用已有的 Remotion 项目
cd my-video-project && npm i

一次性成本： 全新的 Remotion 项目需下载约 2.2 GB npm 包 + 约 90 MB Chrome headless shell。为下一个视频优先复用已存在的 Remotion 项目（已装好 node_modules/）——重安装在每个项目只发生一次，而非每支视频。Lottie 动画为可选（@remotion/lottie + lottie-web），只有用到 LottieAnimation 时才按项目安装。

3. 配置 — 设置 TTS_BACKEND 及其 API 密钥（见 TTS 后端 和 环境变量）。

4. 告诉你的 agent：

"帮我制作一个关于 [你的主题] 的视频播客"

agent 会自动跑完整个流程（研究 → 脚本 → TTS → Remotion 合成 → Studio 预览 → 4K 渲染 + BGM）。渲染前可在 Remotion Studio（npx remotion studio src/remotion/index.ts）预览迭代；agent 会等你说出明确的"渲染 4K"才进行最终渲染。

⚠️ 给读到这里的你（不是给 AI 看的）：podcast.txt 必须人工反复打磨

这一节是写给你这个真人的，不是写给 agent 的。 整条流水线后面所有环节 — TTS 朗读、字幕、章节切换、动画节奏、最终成片 — 全部由这一份 podcast.txt 决定。脚本不行，4K 渲出来的也是垃圾。

AI 生成的初稿只是起点。请你亲自做下面这些事，不要交给 AI 代劳：

1. 按口播节奏在脑子里默读。 每句话当成一口气说完 — 哪句让你"换气换不过来"、哪句要回头重读才懂，立刻改。读得舒服 ≠ 听得舒服，TTS 卡住的地方往往就是你默读时也卡的地方。
2. 至少改三遍。
   • 第一遍：抓错别字、明显语病、绕口处
   • 第二遍：砍废话、砍套话、砍"那么我们今天就来聊一聊"这种开场
   • 第三遍：调节奏 — 哪里断句、哪里加停顿、长句切短、重音落在哪个词上
3. 逐章节通读。 每个 [SECTION:xxx] 块从头看到尾，确认开头有钩子、结尾能自然过渡到下一节，不是一堆并列要点堆在一起。
4. 数字 / 专有名词 / 英文术语单独审一遍。 TTS 念错的 90% 都集中在这里。读音不对的，去 phonemes.json 加词条；读着别扭的，直接换说法。
5. 心里要有时长账。 中文按 每分钟约 280 字 估算（英文约每分钟 150 词）。目标 5-10 分钟 ≈ 1400-2800 字，不要凑。

校验通过的唯一标准：脑子里走完一遍，没有任何一句让你皱眉。 达不到就不要进 Step 7（TTS），否则你只是在用 4K 渲染一段连你自己都不想听完的内容。

工作流程

相关技能

本仓库内的变体（skills/）：

• video-podcast-maker —— 完整制作管线（本 README 的主角）
• video-podcast-maker-lite —— 极简个人管线：Azure SSML TTS + Remotion，不带模板
• video-podcast-maker-nano —— 工具无关的纯逻辑管线（任意 TTS 后端、任意视频工具）；默认全自动，审查级别按项目配置

外部技能：

• remotion-best-practices - 推荐；Remotion 核心模式与规范（缺失时使用内置最简规则）
• assetseeker - 可选；许可核验的图库/视频/BGM/音效/图标/字体
• imagencn - 可选；AI 图片与封面（付费 API）
• videogencn - 可选；AI 视频片段，用于 B-roll（付费 API）
• Hyperframes - 可选；透明动画叠层（Node 22+）

环境要求

软件	版本	用途
macOS / Linux	-	已在 macOS 测试，兼容 Linux
Python	3.8+	TTS 脚本、自动化
Node.js	18+	Remotion 视频渲染
FFmpeg	4.0+	音视频处理

通过 skills CLI 安装的？ 届时 SKILL.md / scripts / templates 位于 agent 暴露的 ${SKILL_DIR} 路径下；README 中的路径写法是仓库根目录视角，也就是克隆安装后的样子。

TTS 后端（本地）

TTS 内置合成，无需外部组件技能。TTS_BACKEND 直接填平台 id，只需配置当前平台的环境变量：

TTS_BACKEND	平台	所需环境变量	获取密钥
edge（默认）	微软 Edge TTS	（无 —— 免费）	—
azure	微软 Azure Speech	AZURE_SPEECH_KEY、AZURE_SPEECH_REGION（默认 eastasia）	Azure 门户

想要更多平台？ 原先的 ttscn 组件技能（cosyvoice、豆包、腾讯云、百度、MiniMax、讯飞、ElevenLabs、OpenAI、Google）不再是依赖。如需这些平台，请单独安装并直接调用。

环境变量

添加到 ~/.zshrc 或 ~/.bashrc：

export TTS_BACKEND="edge"                  # edge（默认）/ azure
export TTS_VOICE="zh-CN-XiaoxiaoNeural"    # 可选；不设置则用后端默认音色
export TTS_RATE="+5%"                      # 可选；也可写入 user_prefs.json（global.tts.rate）
export TTS_STYLE="gentle"                  # 可选；仅 azure 生效
export AZURE_SPEECH_KEY="..."              # 仅 azure 需要（见上表）
export AZURE_SPEECH_REGION="eastasia"      # azure 语音区域
export GEMINI_API_KEY="..."                # 可选：AI 封面（imagencn）
export DASHSCOPE_API_KEY="..."             # 可选：AI 封面（imagencn；ark/混元/智谱/step 亦可）

然后重新加载：source ~/.zshrc

配置文件

可变用户级文件位于 ~/.video-podcast-maker/（所有项目共享，技能更新不会覆盖）；其余文件位于技能根目录（本仓库为 skills/video-podcast-maker/，通过 skills CLI 安装后为 ${SKILL_DIR}）：

文件	位置	说明
phonemes.json	~/.video-podcast-maker/	全局多音字词典；首次运行自动从内置模板复制；项目级覆盖放 videos/{名称}/phonemes.json
user_prefs.json	~/.video-podcast-maker/	你的偏好（TTS、BGM、平台、视觉覆盖、风格档案）；首次运行自动从模板复制
user_prefs.template.json / phonemes.template.json	技能根目录	默认模板 — 用户级副本的来源
prefs_schema.json	技能根目录	偏好校验的 JSON Schema
tsconfig.json	技能根目录	Remotion 模板的 TypeScript 配置

输出结构 — 每个视频渲染到自己的 videos/{名称}/ 目录：

videos/{视频名称}/
├── topic_definition.md      # 主题定义
├── topic_research.md        # 研究笔记
├── podcast.txt              # 旁白脚本
├── phonemes.json            # （可选）发音覆盖
├── assets/manifest.json     # 资产清单（角色 / 来源 / 许可）
├── podcast_audio.wav        # TTS 音频
├── podcast_audio.srt        # 字幕文件
├── timing.json              # 章节时间轴（驱动动画同步）
├── thumbnail_*.png          # 视频封面
├── publish_info.md          # 标题、标签、简介
├── output.mp4               # 4K 原始渲染
├── video_with_bgm.mp4       # 含背景音乐
├── bgm.mp3                  # 背景音乐
├── final_video.mp4          # 最终输出
└── shorts/                  # （可选）9:16 竖屏短片

背景音乐： 内置曲目位于 skills/video-podcast-maker/assets/ —— perfect-beauty-191271.mp3（轻快积极）和 snow-stevekaldes-piano-397491.mp3（舒缓钢琴）。各平台行为（封面、章节、CTA、发布格式）见技能内 references/platform-matrix.md。

❤️ 支持作者

如果这个项目对你有帮助，欢迎支持作者：

微信支付

支付宝

Buy Me a Coffee

打赏

💬 支持

如有问题、功能建议或疑问，请在 GitHub 提交 Issue。

👤 作者

Agents365-ai

• B站: https://space.bilibili.com/441831884
• GitHub: https://github.com/Agents365-ai

📄 开源协议

MIT — 免费使用、修改、再分发，只要包含版权与许可声明。