mcps
E R D

Recipe

Turn a script into a narrated video

Read your script aloud in ElevenLabs, generate matching scenes in Runway, and assemble a narrated explainer — from text, with no footage.

for Anyone who wants a short video from a photo or a sentence — no editing software

Hand-checked recipe · curated by Ilya Gindin

~45 minutes

to run

intermediate

difficulty

3

connectors

What you'll need

ElevenLabs Reads your script aloud in a chosen voice — the narration the video is built around.
Required
Runway Generates the cinematic scenes that match each beat of the script.
Required
Descript Optional — lays the narration over the scenes, adds captions, and exports the final video.
Optional

This recipe runs in 3 steps

  1. 1

    Narrate the script in ElevenLabs

    Say to your AI

    Narrate this script in a calm, documentary voice with natural pauses between sentences: '[paste your script]'.

    You'll get: ElevenLabs returns the full narration as one audio file.

  2. 2

    Generate the scenes in Runway

    Say to your AI

    Generate four 4-second cinematic scenes, 16:9, realistic with consistent warm lighting, one for each beat: [describe beat 1], [beat 2], [beat 3], [beat 4].

    You'll get: Runway returns short clips, one per beat of the script.

  3. 3

    Assemble the video

    Say to your AI

    Lay the narration over the Runway scenes in order, add captions that match the words, and export a 1080p MP4.

    You'll get: A finished narrated explainer. (No Descript? Drop the clips and the audio into any free editor — they're already timed to the script.)

You're done when

A short narrated explainer video — your script read in a natural voice over matching cinematic scenes — built from text in well under an hour, with no footage and no editor.

The chain is ElevenLabs → Runway (+ optionally Descript), about 45 minutes: narrate the script first, size each generated scene to the sentence it covers, and repeat the same style words across every shot. That order — voice first, scenes second — is what keeps the film in sync.

An explainer video usually means a script, a voice actor, footage, and an editor. This recipe replaces the last three with two connectors that don’t know about each other: ElevenLabs reads the script, Runway paints the scenes, and you lay them together. No single repo or app-store page covers that hand-off — each tool only documents itself.

The order matters, and it’s the part nobody writes down: narrate first, then build each scene to the length of the line it sits under, and carry the same style words across every Runway prompt so the shots feel like one film. Get that sequence right and a written script becomes a watchable video in an afternoon.

Checked 2026-06-14: ElevenLabs and Runway connectors both reachable. Both are official — ElevenLabs installs with your API key; Runway is a local build (clone + npm run build) with your RUNWAYML_API_SECRET.

If something breaks

I don't see Runway to turn on
Fix: Runway's connector is official but it's a local build: clone runwayml/runway-api-mcp-server, run npm install and npm run build, then add it with your RUNWAYML_API_SECRET. It's not a one-click hosted sign-in.
The narration and the video drift out of sync
Fix: Generate the narration first, then size each Runway scene to the chunk of script it covers. Matching scene length to its sentence is what keeps them aligned.
Runway scenes look inconsistent shot to shot
Fix: Repeat the same style words — lighting, lens, palette — in every scene prompt. Runway treats each clip independently, so the visual consistency has to come from you.
It ran out of credits
Fix: Runway renders from your plan's credits and ElevenLabs from your character quota. Generate at lower resolution while you lock the script, then re-render only the keepers.

People ask their AI

“turn a script into a narrated video”“make an explainer video from text”“voiceover plus AI video from my script”

Related

ElevenLabs Runway turn a photo into a narrated reel lo fi track with a looping visual