Skip to content

API changelog

What changed in the API and when.

2026-11-16

n8n node and MCP registry

  • New: the n8n community node n8n-nodes-storkai (version 0.1.1) makes talking photo videos, voiceovers and singing videos inside a workflow.
  • The MCP server is listed in the official MCP registry as ai.stork/mcp.

2026-11-16

Python library

  • New: pip install storkai (version 0.1.0) is a client for the Talking Photo, Speech and Singing Video APIs. No dependencies, Python 3.8 or newer.

2026-11-16

Node library

  • New: npm install storkai (version 2.1.0) is a client for the Talking Photo, Speech and Singing Video APIs, with TypeScript types and webhook verification. The MCP server in the same package is unchanged.

2026-11-16

Singing Video API price

  • The price is now 20 credits per second of delivered video. It was 10 at launch, earlier the same day. POST /singing-video/estimate returns the current figure in credits_per_second.
  • A video may now render for up to 75 minutes before it is marked failed and its credits are returned.

2026-11-16

Singing Video API

  • New: POST /singing-video turns one photo and a song into a performance video, GET /singing-video returns one video or lists recent ones, POST /singing-video/estimate prices a length.
  • Signed webhooks with the event types singing_video.succeeded and singing_video.failed.
  • The same calls as MCP tools: stork_singing_video_create, _status and _estimate.

2026-11-16

Webhooks for the Speech API

  • POST /speech takes webhook_url. We POST once when the file is done or has failed, signed like the other webhooks; event types speech.done and speech.failed.

2026-11-16

api.stork.ai

  • The API has its own host: https://api.stork.ai. mcp.stork.ai serves the same routes and keeps working, so nothing needs to change in an existing integration.
  • GET https://api.stork.ai/ returns an index of the APIs, their endpoints and the docs.

2026-11-15

Speech API

  • New: POST /speech turns text into an MP3, GET /speech returns one file or lists recent ones, POST /speech/estimate prices a text without making anything.
  • Same key and credits as the Talking Photo API. A free tier with "tier": "free".
  • The speech endpoints are in openapi.json.

2026-11-15

German, Japanese and Korean voices, and your own voice

  • Your own voice: a voice added to your account in the dashboard works as a voice_id.
  • Six more built-in voices. GET /voices lists them; the ids start with de-, ja- and ko-.

2026-11-14

Talking Photo API v1.1

  • Text instead of audio: send text and a voice_id and a built-in voice reads it. GET /voices lists the voices.
  • Free tier: "tier": "free" makes a watermarked video without spending credits.
  • Safe retries: an Idempotency-Key header replays the first response and never charges twice.
  • Signed webhooks: pass webhook_url; verify Stork-Signature with your signing secret.
  • New: POST /talking-photo/estimate and GET /talking-photo?limit=.
  • An OpenAPI 3.1 spec at /api/v1/openapi.json, and the same calls as tools on the MCP server.
  • A 402 insufficient_credits starts an auto-recharge when it is switched on, so a retry a minute later succeeds.
  • Links to paid videos last 30 days.

2026-11-13

Talking Photo API v1

  • New: POST /talking-photo with image_url and audio_url, and GET /talking-photo?id= for the result.
  • aspect_ratio (original, 16:9, 9:16, 1:1) and resolution (480p, 720p, 1080p).
  • Paid from prepaid credits, billed per started 10 seconds of finished video.

Docs: Talking Photo API · Speech API