You can now create voiceovers using Gemini as well as ElevenLabs. That means a wider range of voices and delivery styles to choose from when you're bringing a Character to life or generating spoken audio inside Imaginations.
Both engines live side by side, so you can pick the one that best fits the tone, language, or pacing your project calls for, all from within your brand-aware workspace.
What's new: voiceovers inside Flows
You can now generate voiceovers directly from a Flow. An audio node can be switched into text-to-speech mode, which lets you turn any text in your Flow into spoken audio using either ElevenLabs or Gemini, without leaving the canvas.
How to use it
Open a Flow and add an audio node from the node toolbar.
Set the node to text-to-speech mode.
Click the voice picker on the node and search for the voice you want. Voices are grouped under a dedicated speech category so they're easier to scan.
If you choose a Gemini voice, use the extra style and pace controls to shape the delivery.
Connect your text input to the node and run the Flow to generate the voiceover.
A couple of small quality-of-life improvements landed alongside this:
The voice picker is searchable, so it's quick to find a specific voice in long lists.
The node toolbar now includes canvas zoom controls, making it easier to work on larger Flows.
If you try to use a text-to-speech voice on a non-TTS audio node, you'll see a friendly error explaining that the node needs to be in text-to-speech mode. Just switch the node's mode and try again.
How to use it with Characters
Open a Character (or create a new one) from your workspace.
In the Character form, find the voice model selector and choose either ElevenLabs or Gemini.
Pick a specific voice from the available options for that engine.
Adjust the delivery controls to shape how the voice performs — for example, pacing or tone.
Save the Character. Any voiceover you generate with this Character will now use the engine and voice you selected.
You can also choose the voice engine directly when creating a one-off voiceover, without attaching it to a Character.
Which one should I use?
There's no single right answer — it depends on the feel you're after:
ElevenLabs remains a strong choice for highly expressive, character-driven performances and voice cloning workflows.
Gemini brings a different set of voices and delivery controls, which can be useful when you want a different tonal range or a fresh option to A/B against.
Since both engines are available in the same place, the easiest approach is to generate a short test line with each and compare them against your brand voice guidelines.
Tips
If you change the voice model on an existing Character, new voiceovers will use the updated engine, but previously generated audio files won't be re-rendered automatically.
Voice availability and supported languages differ between ElevenLabs and Gemini, so check both lists if you don't see the voice you expect.
Voiceovers count toward your workspace usage. Keep an eye on your plan if you're generating long-form audio.
For more on building voice-enabled Characters, see Characters.