Skip to main content

Recipe: Text to speech

Goal: turn a script into spoken audio — for narration, voiceovers, or accessibility.

Nodes you'll use: Text Input → a TTS node (ChatterboxTTS, or the OpenVoice nodes for voice cloning/conversion) → an audio output.

In the Workflow Builder​

  1. Add a Text Input node and paste your script (or wire it to a file).
  2. Add a ChatterboxTTS node.
  3. Connect Text Input → ChatterboxTTS, Save, and Run.
  4. Download the generated audio.

Want a specific voice? The OpenVoice nodes do voice cloning / conversion — feed a reference clip plus your text.

Pass the text at runtime (API)​

The text doesn't have to be baked into the node — pass it per-run with extra_variables:

# $AUTH is "Authorization: Bearer <token>" — get it from your sk_ key,
# see /api-reference#authentication
curl -s -X POST "$BASE/workflows/workflow-versions/<your_version_id>/run" -H "$AUTH" \
  -H "Content-Type: application/json" \
  -d '{"cog_ids": [], "extra_variables": {"text_content": "Welcome to MOD. Let'\''s build something."}}'

The extra_variables keys are workflow-specific. Call GET /workflows/workflow-versions/{id}/variables to see which keys your workflow accepts — a Text Input typically exposes text_content. Passing it overrides the node at run time, handy for generating many clips from one workflow. See the API Reference.