MCP 服务器

@scriptivox/mcp-server

Enables AI assistants to transcribe audio and video from URLs or local files with high accuracy, speaker diarization, 119 languages, and word-level timestamps, while also supporting transcription management and caption export in SRT, WebVTT, or plain text.

README

@scriptivox/mcp-server

MCP (Model Context Protocol) server for Scriptivox — AI-powered audio and video transcription.

Turn any AI assistant into a transcription powerhouse. Transcribe audio and video from URLs or local files with 99% accuracy, speaker diarization, 119 languages, and word-level timestamps. Plus full CRUD on transcriptions (cancel, delete, list) and caption export in SRT / WebVTT / plain text.

Quick Start

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "scriptivox": {
      "command": "npx",
      "args": ["-y", "@scriptivox/mcp-server"],
      "env": {
        "SCRIPTIVOX_API_KEY": "sk_live_YOUR_KEY"
      }
    }
  }
}

Claude Code

claude mcp add scriptivox -- npx -y @scriptivox/mcp-server

Then set the environment variable SCRIPTIVOX_API_KEY=sk_live_YOUR_KEY.

Other MCP Clients

Any MCP-compatible client can use this server via stdio transport:

SCRIPTIVOX_API_KEY=sk_live_YOUR_KEY npx -y @scriptivox/mcp-server

Getting an API Key

Sign up at platform.scriptivox.com
Go to Keys in the dashboard
Create an API key (starts with sk_live_...)
Add credits at Billing ($5 minimum — API pricing is $0.20/hour of audio)

Tools

Discovery tools (no API key required)

Tool	Description
`get_supported_languages`	List all 119 supported transcription languages with ISO codes
`get_pricing`	View plans, API pricing, and per-file limits
`get_product_info`	Learn about Scriptivox features (`transcription`, `audio-tools`, `video-tools`, `subtitle-tools`, `meeting-bot`, `api`, `all`)
`get_api_docs`	API documentation sections (`quickstart`, `transcribe`, `result`, `list`, `cancel`, `delete`, `upload`, `balance`, `webhooks`, `errors`, `all`)

Transcription tools (API key required)

Tool	Description
`transcribe_url`	Transcribe audio/video from a public URL (Google Drive, Dropbox, OneDrive, or direct file URLs). Supports `language`, `diarize`, `speaker_count`, `align`, `webhook_url`, `idempotency_key`, `await_completed`.
`transcribe_upload`	Transcribe a LOCAL file. Drives the 3-step upload flow internally. Up to 5 GB.
`transcribe_status`	Check the status of a transcription by ID. Returns the full transcript when completed.
`transcribe_cancel`	Cancel an in-flight transcription. Refunds reserved balance. Idempotent.
`transcribe_delete`	Soft-delete a transcription record. Idempotent. Refuses to delete in-flight jobs.
`list_transcriptions`	List recent transcriptions with `status`, `from`, `to`, `limit`, `cursor`, `order` filters.
`export_transcript`	Export a completed transcript as SRT subtitles, WebVTT subtitles, or plain text. Segmentation knobs: `max_words`, `max_chars`, `max_duration`, `sentence_aware`, `include_speakers`, `strip_chars`.
`check_balance`	View your API credit balance and estimated hours available.

Tip: always pass `language` when you know it

Auto-detection works in most cases but has a small failure rate on short clips, code-switched audio, or files starting with music. Passing the ISO code is both faster and more accurate.