video-caption-generator

To automatically generate social captions and YouTube titles from Google Drive videos, you can use a workflow that scans your Drive folder, transcribes the new video files locally, and passes the text to an AI model to produce platform-ready marketing copy.

Video deduplication works by computing a content hash from the video transcript. This identifies clips with identical audio, allowing the workflow to skip reprocessing while still preserving A/B filename variants for separate tracking.

Yes, you need an Anthropic API key for Claude to generate conversational social captions and curiosity-driven titles. You also need local Whisper transcription software to convert the video audio into text first.

Yes, you can use local Whisper transcription for short-form videos stored in Google Drive. The workflow downloads MP4 or MOV files from a specified folder ID and processes them locally using the Whisper turbo model.

The best way to handle A/B testing variants when automating video transcription is to deduplicate clips by their transcript-derived content hash while preserving the unique filenames, ensuring each variant is tagged correctly without reprocessing identical audio.

Videos are reprocessed if persistent tracking of processed Drive file IDs is missing or cleared. The workflow requires persistent tracking of these IDs to support reruns without reprocessing previously handled files.