Turn a reel script into talking character videos
Split a script into lines, voice each one with ElevenLabs and make a lip-synced character clip per line with Kling on fal.ai
Version 1
What it does
Part of the AI character reel system. An Airtable button on a script record calls the webhook. A Code step splits the script into one item per spoken line and labels each with its line number and speaker, and a generated asset record is created for every line.
For each line the workflow finds the speaker's approved character profile, turns the line into speech with that character's ElevenLabs voice, and uploads the audio to Cloudinary. It then sends the character image and the audio to the Kling AI avatar model on fal.ai, checks every 10 seconds until the clip is ready, and saves the video file, line text, character and duration to the asset record.