Video Studio

The Video Studio is River's video production surface. Ingest video files, run AI analysis, edit clips, generate new video with cloud or local AI providers, and ship final assets — all inside the same workspace as your scripts, transcripts, and creative ops.

Pipeline at a glance#

1. Ingest — drop in a video, or import from a connected service. River extracts audio, runs transcription, generates a timeline, and stores the result as a Video entity.
2. Analyze — run AI over the transcript and frames: scene segmentation, speaker detection, content tagging.
3. Edit — non-linear editing inside River with multi-track timeline, trims, overlays, captions.
4. Generate — text-to-video and video-to-video with Replicate, FAL, Runway, Luma, Topaz.
5. Ship — export as MP4 or push to a publishing Workflow.

Built-in transcription#

Every video gets a transcript with speaker detection. The transcript is itself a Transcript entity, fully searchable and AI-readable. Skim the transcript to navigate the video; click a line to jump to that moment.

AI generation#

Video Studio integrates with the major video AI providers:

Set your provider keys in Settings → AI → Video Providers.

What people build#

• Short-form content pipelines — script → AI generation → edit → publish, mostly automated
• Educational content — long video → AI extracts highlights → publishes as clips
• Product demos — script in a Doc, generate scene-by-scene, edit together
• Social cuts — long meeting → AI picks 3 best minutes → social-ready clips

ffmpeg requirement#

Video Studio needs ffmpeg and ffprobe. They're installed automatically with River; for self-hosted setups, see the deployment guide.

Tips#

• Start with the transcript. It's where most of the AI value lives.
• Pair Video Studio with a publishing Workflow — generate, edit, post, all on a schedule.
• For raw cost reasons, generation often beats stock footage for variety; expect this gap to widen.