extracting-transcripts

Transcribe video to text with word-level timestamps using GPU-accelerated WhisperX on local media files, which outputs JSON containing precise word timings suitable for caption and search workflows.

Transcribe audio from a URL without a local GPU by utilizing the IX Toolkit API for cloud transcription, processing long-form media remotely and returning text or SRT caption formats instead of JSON.

This speech-to-text method supports SRT caption output alongside JSON with word timings, allowing direct generation of standard subtitle files for video content and podcasts.

Transcribing media for captions and search outputs JSON with word-level timing and standard SRT text formats, supporting both detailed Remotion-based workflows and direct subtitle application.

You need a local GPU to run WhisperX transcription for local media files, ensuring high accuracy and fast turnaround through GPU acceleration, though cloud transcription via the IX Toolkit API is also available.