Narration
generate spoken audio from text
Before you start
- The text to speak.
- An available voice and the desired language.
How to use this workflow
Step 1
Open Voice and prepare the narration text.
Step 2
Choose a voice/model and review the pronunciation and delivery controls available for it.
Step 3
Listen to the complete output before using it in a project.
What to check
Review names, numbers, emphasis, and pauses; delivery controls differ by voice model.
Prepare the inputs
- Finalize the script’s meaning before generation. Check names, numbers, dates and abbreviations and mark the intended pauses with clear punctuation.
- Choose the target language and an authorized available voice. Decide the delivery style and the duration the surrounding project can accommodate.
Choose the right controls
Open Voice and inspect the selected model’s language and delivery controls. Use the saved voice when continuity matters, then check the available settings rather than assuming every model exposes the same controls.
Troubleshoot the result
Names, numbers or emphasis are wrong
Revise the script or the pronunciation controls actually offered by the selected model. Compare the new audio against the written script from beginning to end, including the final word.
The voice is right but the pacing is not
Simplify long sentences and clarify pauses. Review the narration against the intended picture or timeline; a saved voice does not guarantee the same delivery on a different script.
Continue with an approved result
Keep the approved audio for Talking Head or Direct. For another language, review a translated script and language support before generating; do not infer translation accuracy from natural-sounding speech.
Model controls to review
Related workflows
Sources and basis
- ElevenLabs v3 inputs and settings in DayGen — DayGen product information; accessed