How do you use speech-to-text to direct AI agents faster during video production?

Enable your operating system's dictation (or any speech-to-text app) and speak your direction straight into the invideo agent's chat in plain on-set language — 'stay on him, no cutting, hold until he lunges.' Spoken input keeps you thinking in shots instead of prompt syntax, which matters most when you're running 6–8 sub-agents in parallel.

Set up voice input where you already direct. invideo is an agentic video creation tool — direction goes in as natural language chat, so speech-to-text needs no special integration: turn on built-in OS dictation (macOS or Windows) or a dedicated dictation app, put your cursor in the invideo agent's chat, and speak. Anything you would have typed, you can now say, and you stay in the shot you're imagining instead of in a text box.

Speak the way you'd talk to a crew, not the way you'd write a prompt. Voice input works because conversational directing is already the recommended posture — give directorial intent, not technical parameters. A documented production directed shots with lines like 'I want to stay on the feral guy when we run this scene. No back and forth cutting. We hold on him right up till he lunges,' and the invideo agent executed exactly that intent. Spoken sentences naturally come out in this register; typed prompts tend to collapse into parameter lists.

Route spoken direction across your sub-agent crew. Initialize a creative producer agent with the full script, shot breakdown, and character details first — it holds the project's vision — then dictate shot notes to a DOP agent, feedback to a storyboard agent, costume choices to a costume designer agent. This is where speech-to-text pays off most: documented productions ran 6 agents simultaneously in one setup and 8 specialist agents across separate project pages in another, and typing into that many threads is the bottleneck voice removes. One creator credits AI-agent workflows with making the pipeline 5x faster; a 2-minute brand promo built this way finished in 3 days versus an estimated week of manual prompting.

Let locked context keep your spoken commands short. Because the invideo agent holds project context persistently, brief spoken instructions resolve correctly — in one production, a three-word continuation ('Everything should match') was enough to carry character, lighting, and spatial continuity into the next sequence. The shorter your commands can safely be, the faster dictation gets.

Fix transcription errors on proper nouns before they propagate. Speech-to-text reliably stumbles on character names and technical film vocabulary, so glance at the transcribed text before sending, and keep character names spelled consistently in the loaded script so a spoken reference still maps to the right character sheet. The invideo agent also surfaces ambiguity rather than guessing — it asks clarifying questions when a direction has a gap — which catches most mistranscriptions before credits are spent.

Type only where precision beats speed. Exact spec entries and manual overrides — a precise re-crop, a specific corrected value you'll log back to the invideo agent's shot breakdown — are faster typed. Dictate the directing; type the data.

Watch some of these to see what works for you:

I wanna talk about my shot like this because then I can keep thinking about my entire film in my head without breaking it.

— invideo's creative team