scribiz

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Scribiz

Scribiz turns a video or audio link, or a local file, into a transcript with speakers and timestamps, the text shown on screen, a summary, and chapters. It works when a video has no captions.

Docs: scribiz.com/docs. Every docs page is also Markdown: add

.md

to the path.

Scribiz 可将视频或音频链接，或本地文件，转换为包含说话者和时间戳的文字稿、屏幕上显示的文本、摘要和章节。即使视频没有字幕也可使用。

文档：scribiz.com/docs。每个文档页面也提供 Markdown 版本：在路径末尾添加

.md

。

Pick a way in

选择接入方式

What does the agent have?
├── An MCP client (Claude Code, Cursor, Codex) and a link
│   └── Use the MCP server. Best for questions, summaries and search.
├── A shell and a link or local file
│   └── Use the CLI. Best for subtitles, files on disk and batch jobs.
└── Neither
    └── Send the user to scribiz.com to paste the link.

What does the agent have?
├── An MCP client (Claude Code, Cursor, Codex) and a link
│   └── Use the MCP server. Best for questions, summaries and search.
├── A shell and a link or local file
│   └── Use the CLI. Best for subtitles, files on disk and batch jobs.
└── Neither
    └── Send the user to scribiz.com to paste the link.

MCP server

Hosted at

https://scribiz.com/mcp

(Streamable HTTP). No key needed to start.

bash

undefined

托管于

https://scribiz.com/mcp

（Streamable HTTP）。启动无需 API key。

bash

undefined

Claude Code

claude mcp add --transport http scribiz https://scribiz.com/mcp


Cursor, in `~/.cursor/mcp.json`:

```json
{ "mcpServers": { "scribiz": { "url": "https://scribiz.com/mcp" } } }

toml

undefined

claude mcp add --transport http scribiz https://scribiz.com/mcp


Cursor，配置文件为 `~/.cursor/mcp.json`：

```json
{ "mcpServers": { "scribiz": { "url": "https://scribiz.com/mcp" } } }

toml

undefined

Codex: ~/.codex/config.toml

[mcp_servers.scribiz] url = "https://scribiz.com/mcp"


Tools:

- `get_video_context`: summary, chapters and key moments in under 2,000 tokens. Start here.
- `search_video`: find where a phrase is said and read only those minutes.
- `ask_video`: an answer with 3–5 cited moments, each a timestamp link.
- `get_transcript`: the full transcript, when you really need all of it.
- `get_job`: check a running job.

Without a key the server reads captions, videos already processed and, within a small daily allowance, the link itself. It never looks at the picture. With an API key (`Authorization: Bearer $SCRIBIZ_API_KEY`) it uses the account's minutes and can read the screen. Other clients: [MCP install](https://scribiz.com/docs/mcp/install.md).

Prefer `get_video_context` then `search_video` over pulling the whole transcript. A two-hour talk is more than 25,000 tokens.

[mcp_servers.scribiz] url = "https://scribiz.com/mcp"


工具：

- `get_video_context`：以不到 2,000 tokens 的篇幅提供摘要、章节和关键时刻。请先使用此工具。
- `search_video`：查找某个短语出现的位置，并只读取对应片段。
- `ask_video`：返回带有 3–5 个引用时刻的答案，每个引用均为时间戳链接。
- `get_transcript`：获取完整文字稿，仅在确实需要全文时使用。
- `get_job`：检查正在运行的任务。

没有 API key 时，服务器会读取字幕和已处理的视频；在较小的每日额度内，还可直接读取链接本身。该模式不会查看画面。配置 API key（`Authorization: Bearer $SCRIBIZ_API_KEY`）后，可使用账户的分钟数配额并读取屏幕内容。其他客户端：[MCP 安装](https://scribiz.com/docs/mcp/install.md)。

优先使用 `get_video_context`，然后使用 `search_video`，而不是拉取完整文字稿。一场两小时的演讲会超过 25,000 tokens。

CLI

bash

npm install -g scribiz     # or: npx scribiz --help
scribiz login              # free account, 30 minutes a month
scribiz setup              # or: your own Gemini API key
scribiz doctor             # checks the credential, ffmpeg, ffprobe, yt-dlp

Needs Node.js 24+,

ffmpeg

, and

yt-dlp

for most links (

brew install ffmpeg yt-dlp

). Tested on macOS.

bash

scribiz "https://www.youtube.com/watch?v=VIDEO_ID" -o talk.srt        # subtitles
scribiz talk.mp4 -f txt -o talk.txt                                    # plain text
scribiz talk.mp4 -f md -o talk.md                                      # Markdown with speakers
scribiz context talk.mp4 --visual -f context                           # transcript + on-screen notes + summary + chapters, for an LLM
scribiz talk.mp4 --json --out-file talk.json                           # save the full result once
scribiz format talk.json -f vtt                                        # render another format without paying again
scribiz ask talk.json "What did they decide about pricing?"

Put links that contain

in quotes. A folder input writes one

.srt

next to each file.

bash

npm install -g scribiz     # or: npx scribiz --help
scribiz login              # free account, 30 minutes a month
scribiz setup              # or: your own Gemini API key
scribiz doctor             # checks the credential, ffmpeg, ffprobe, yt-dlp

处理大多数链接需要 Node.js 24+、

ffmpeg

和

yt-dlp

（

brew install ffmpeg yt-dlp

）。已在 macOS 上测试。

bash

scribiz "https://www.youtube.com/watch?v=VIDEO_ID" -o talk.srt        # subtitles
scribiz talk.mp4 -f txt -o talk.txt                                    # plain text
scribiz talk.mp4 -f md -o talk.md                                      # Markdown with speakers
scribiz context talk.mp4 --visual -f context                           # transcript + on-screen notes + summary + chapters, for an LLM
scribiz talk.mp4 --json --out-file talk.json                           # save the full result once
scribiz format talk.json -f vtt                                        # render another format without paying again
scribiz ask talk.json "What did they decide about pricing?"

将包含

的链接放在引号内。输入文件夹时，会在每个文件旁写入一个

.srt

文件。

Options

选项

Format:
```
-f srt|vtt|txt|md|context|json
```
(default
```
srt
```
)
Mode:
```
-m auto|captions|audio|visual|full
```
(aliases
```
listen
```
,
```
watch
```
,
```
both
```
). Auto uses good captions and listens when there are none.
Speakers:
```
--speakers
```
for interviews and podcasts,
```
--no-speakers
```
for on-screen captions
Proofread:
```
--proofread
```
, plus
```
--vocabulary "Brand,Name"
```
for names speech-to-text mishears
Language hint:
```
-l en
```
Cost guard:
```
--max-cost 0.50
```
,
```
--max-duration 02:00:00
```
Scripts:
```
--no-input
```
never prompts. Exit codes: 0 ok, 2 usage, 3 sign-in or quota, 4 source, 5 provider, 6 missing tool.

Do not split or rechunk SRT cues after the fact. Re-run Scribiz instead.

格式：
```
-f srt|vtt|txt|md|context|json
```
（默认值为
```
srt
```
）
模式：
```
-m auto|captions|audio|visual|full
```
（别名为
```
listen
```
、
```
watch
```
、
```
both
```
）。Auto 会使用质量较好的字幕，没有字幕时则进行音频转录。
说话者： 访谈和播客使用
```
--speakers
```
，屏幕字幕使用
```
--no-speakers
```
。
校对： 使用
```
--proofread
```
，并通过
```
--vocabulary "Brand,Name"
```
修正 speech-to-text 误识别的姓名。
语言提示：
```
-l en
```
成本限制：
```
--max-cost 0.50
```
、
```
--max-duration 02:00:00
```
脚本：
```
--no-input
```
可避免交互提示。退出码：0 表示成功，2 表示用法错误，3 表示未登录或额度不足，4 表示来源错误，5 表示提供商错误，6 表示缺少工具。

不要在事后拆分或重新分块 SRT cue。请改为重新运行 Scribiz。

Content workflows

内容工作流

Video to blog post: run
```
get_video_context
```
(or
```
scribiz context
```
), outline from the chapters, quote the transcript for key claims, and link timestamps as sources.
Podcast show notes:
```
-f md --speakers
```
, then write a summary, chapter list with timestamps, and pull quotes.
Social clips: use
```
search_video
```
to find the strongest moments, then write posts that cite the exact second.
Competitor research: summarize a competitor's demo or webinar and list the claims and features they show on screen (
```
--visual
```
).
Subtitles for upload:
```
--proofread --no-speakers -o video.srt
```
.

视频转博客文章： 运行
```
get_video_context
```
（或
```
scribiz context
```
），根据章节整理提纲，从文字稿中摘录支撑关键观点的原话，并将时间戳链接作为来源。
播客节目说明： 使用
```
-f md --speakers
```
，然后撰写摘要、带时间戳的章节列表和金句摘录。
社交内容： 使用
```
search_video
```
查找最精彩的片段，然后编写注明精确秒数的帖子。
竞品研究： 总结竞品的演示或网络研讨会，并列出其在屏幕上展示的说法和功能（
```
--visual
```
）。
用于上传的字幕：
```
--proofread --no-speakers -o video.srt
```
。

Limits

限制

YouTube, podcast feeds, direct media links and files work best. TikTok, X and Vimeo are best effort. Instagram usually needs
```
--cookies-from-browser chrome
```
on the CLI.
Not for live streams, DRM services (Spotify, Netflix), editing, or generating video.
Spot-check names and brands in the output before publishing.

YouTube、播客 feed、直接媒体链接和文件的效果最佳。TikTok、X 和 Vimeo 尽力支持。在 CLI 中处理 Instagram 通常需要使用
```
--cookies-from-browser chrome
```
。
不支持直播、DRM 服务（Spotify、Netflix）、视频编辑或视频生成。
发布前请抽查输出中的姓名和品牌名称。