Multilingual narration
create speech in supported languages
Before you start
- Reviewed text in the target language.
- A voice/model that supports that language.
How to use this workflow
Step 1
Prepare the translated script before opening Voice.
Step 2
Choose a compatible model and voice, then review available language controls.
Step 3
Have a speaker of the target language review pronunciation, meaning, and pacing.
What to check
Language coverage depends on the selected voice model; a fluent-sounding result can still contain mistakes.
Prepare the inputs
- Have the target-language script reviewed for meaning, tone, names and numbers. Translation and speech generation are separate steps.
- Check whether the selected model and voice support the target language. Account for wording that may take longer to speak than the original.
Choose the right controls
Prepare the reviewed script in Voice and select a compatible voice/model. Use the available language and delivery settings; do not assume support because another model from the same provider accepts the language.
Troubleshoot the result
Pronunciation is plausible but wrong
Ask a speaker of the target language to check the full recording against the script. Correct names, abbreviations and ambiguous wording using the actual controls available, then review the changed audio again.
The translation no longer fits the video
Revise the phrasing or timing while preserving the message. Use Dubbing for its supported source-video workflow when appropriate; raw narration alone does not promise synchronized lips or matching segment lengths.
Continue with an approved result
Save the approved language version with its reviewed script and use it in Talking Head or Direct. Repeat the language review for every version instead of treating one approved language as proof for the others.
Model controls to review
Related workflows
Sources and basis
- ElevenLabs v3 inputs and settings in DayGen — DayGen product information; accessed