Fundamentals -> Template Editor

AI voice-overs

Audio elements can generate speech from text. Instead of uploading an audio file, you write the words to be spoken, and Creatomate sends them to a text-to-speech provider (ElevenLabs or OpenAI) each time the template is rendered. Combined with a dynamic text field, this lets you produce narrated videos entirely through automation.

Creating a voice-over

The quickest way is to choose AI Voiceover from the Audio menu in the toolbar, which video templates have. It adds an audio element that already uses ElevenLabs, with placeholder text for you to replace and the Dynamic switch turned on.

To turn an existing audio element into a voice-over:

  • Select the audio element.
  • In the properties panel, set Provider to ElevenLabs (or OpenAI).
  • Enter the words to be spoken in the Text field.
  • Pick a Model and Voice.

The speech is generated at render time, so you won't hear it while editing. To make the spoken text dynamic, turn on the Dynamic switch at the top of the properties panel, so your automation can pass different words with every render.

Connecting ElevenLabs or OpenAI

The first time you use a provider in a project, you need to connect your own account. The editor shows a notice with an Enable it in Project Settings link, which takes you to the right place. You can also open it manually: go to Project Settings, open the Integration tab, switch on ElevenLabs or OpenAI, and paste the API key from your account with the provider.

Two things to know:

  • The project renders one video at a time while an integration is switched on, including videos that don't use it. This keeps your renders within the provider's rate limits.
  • The provider bills the speech. It is generated on your own account, at the provider's rates, with every render. Creatomate charges its normal render credits and nothing extra.

The OpenAI provider is used the same way as ElevenLabs: choose one of its models (TTS-1 or TTS-1-HD) and voices.

Choosing a model and voice

For ElevenLabs, Eleven Multilingual v2 is the default model and a solid choice for most languages. The Voice menu offers a set of ready-made voices; to use any other voice from your own ElevenLabs account, including cloned voices, select Custom Voice and enter its voice ID from the ElevenLabs dashboard. The Stability, Similarity, Style, and Speaker Boost settings map directly to ElevenLabs' voice settings, so you can fine-tune the delivery without leaving the editor.

Adding subtitles

A voice-over pairs naturally with auto-generated subtitles: point a subtitle element's Source at the voice-over element, and the spoken words appear as synchronized captions. ElevenLabs even reports the exact timing of each word as the speech is generated, so the subtitles match the voice precisely.