Scriptivox for developers

    Scriptivox turns recorded audio and video into text over a REST API, an MCP server and a command-line tool. Everything a developer or an AI agent needs to integrate is linked below, and every one of these URLs is stable.

    The developer platform lives at platform.scriptivox.com. The API itself is at https://api.scriptivox.com/v1.

    Scriptivox API documentation

    The REST API turns recorded audio and video into structured text: 119 languages, speaker diarization, word-level timestamps, and SRT, WebVTT or plain-text export. Files from 1 second to 10 hours, up to 5 GB, in 25 container formats. Billed at $0.20 per hour of audio with no subscription and no minimum commitment.

    • Documentation index
    • Quickstart
    • API reference

    Scriptivox OpenAPI specification

    An OpenAPI 3.1 description of the whole surface: seven operations, each with a unique operationId, a typed input schema and typed responses. Generate a client from it, or diff it in CI to notice a change before it becomes an incident. Every API response also points at it with a Link header carrying rel="service-desc".

    • openapi.json
    • openapi.yaml

    Scriptivox MCP server

    Transcription as native tool calls from Claude, ChatGPT and any other Model Context Protocol client. Run it hosted over Streamable HTTP with nothing to install, or locally over stdio when the media is on your own machine. Four of its tools — pricing, languages, product info and API docs — work with no API key at all.

    • MCP server documentation
    • Manifest
    • Server card (tools + behavioural annotations)
    • Documentation MCP endpoint (search_docs, get_doc)
    • @scriptivox/mcp-server on npm

    Scriptivox CLI

    The official command-line client. Transcribe from a shell or a script, with structured output on stdout, progress on stderr, and exit codes that distinguish a usage error from an API error. Zero runtime dependencies, Node 18 or newer.

    • CLI documentation
    • @scriptivox-api/cli on npm

    Scriptivox API authentication

    Two credentials, for two jobs. An sk_live_ API key — sent as an Authorization or X-Api-Key header — authorises the metered transcription API and suits a script or a server you control. OAuth 2.1 with PKCE is for an agent acting on somebody else’s behalf: the person approves once in a browser and no API key changes hands. Dynamic client registration is deliberately NOT offered — anonymous client self-registration is a surface this deployment does not want — so an OAuth client is onboarded by asking. Discovery starts from any 401, which carries WWW-Authenticate with a link to the protected-resource metadata. Note that an API key and a scriptivox.com web subscription are separate purchases — a subscription grants no API credit.

    • Authentication guide
    • AUTH.MD — both credentials, for machines
    • Protected resource metadata (RFC 9728)
    • Authorization server metadata (RFC 8414) with the agent_auth block
    • Create a key

    Scriptivox webhooks

    HMAC-signed callbacks when a transcription finishes, so a long job does not have to be polled. Three events fire per job, and delivery is fire-and-forget — pair webhooks with a poll as a fallback rather than trusting them alone.

    • Webhooks guide

    Scriptivox API versioning and deprecation policy

    The API is versioned in the URL path at /v1. Breaking changes only ever ship behind a new major version, with six months of notice minimum before anything is removed, signalled on the wire with Deprecation and Sunset response headers rather than only in a blog post.

    • Versioning policy
    • Service status

    Scriptivox for AI agents

    Machine-readable entry points, for an agent deciding whether this service fits its task before it commits to an integration. llms.txt says what Scriptivox is for and — as importantly — when not to use it. Every documentation page has a markdown twin: append .md to its path, or send Accept: text/markdown.

    • llms.txt
    • llms-full.txt (every page, one fetch)
    • This site as markdown
    • Creating an account
    • Paying: every rail, and which ones work
    • Agent skill: the whole loop, in order
    • Skill index
    • Skill index (spec path)
    • Agentic resource catalog (ARD)
    • A2A agent card
    • Pricing, from the live plan catalog
    • Schema.org feed (JSON Lines)
    • Ask a question in natural language (NLWeb)
    • Sitemap

    What Scriptivox is not for

    Live streaming and real-time dictation. Every job takes a complete file. If you need a transcript as someone is still speaking, this is the wrong tool, and knowing that before you integrate is worth more than any of the links above.