Gemini clones your voice with consent: what changes for creators and businesses
Google launched Gemini 3.8 Flash TTS: Gemini clones your voice with consent from 30 seconds of audio. How the consent lock works, the real cost, and whether it replaces ElevenLabs.
On September 23, 2026, Google launched Gemini 3.8 Flash TTS and Flash-Lite TTS. The difference from any earlier voice cloning is that now Gemini clones your voice with consent: it requires a second verbal consent recording before activating cloning, not just the audio sample.
What changed: two models with different purposes
Flash TTS is built for fine creative direction: designing a brand-new voice by describing it in words and directing each line separately (a pause here, a whisper there, an accent shift mid-sentence). Flash-Lite is built for volume: dubbing hours of content or running a customer-service voice agent at a much lower cost per generated minute. Between the two models, Google says it has more than 2,000 production-ready voices, in over 100 languages and dialects.
How the consent lock works
The full flow looks like this: you upload about 30 seconds of audio in AI Studio, and the system asks for a second recording — the voice's owner saying a specific consent phrase out loud. The system compares that second recording against the original sample, and only activates cloning if it matches the same speaker. The generated audio ships with a watermark called SynthID and C2PA credentials a verifier can later read to confirm the audio came from an AI.
The number that changes the dubbing decision
Dubbing a 10-minute video into 5 languages the traditional way means 5 separate studio sessions — using round illustrative numbers, about $100 per session with a professional voice actor already puts you at $500 before exporting anything. With Flash-Lite, you clone the voice once and generate the other four versions as a compute process: minutes, not new sessions. The cost moves from hundreds of dollars per language to cents per thousand words (the API charges per generated character, so it isn't free, but it's a different order of magnitude).
Does this replace ElevenLabs?
The short answer is not yet. ElevenLabs lets you insert very fine performance tags directly into the text, built for long, consistent production episode after episode — it's a production studio's tool. Gemini is betting on something else: that anyone, without a studio account, can clone a voice with verified consent and use it inside tools they already use daily, like Google Vids or a customer-service agent. They aren't competing for the same customer.
Who this doesn't serve yet
If a brand depends on absolute voice consistency episode after episode, switching cold to a new system without testing it first risks the sound its audience already recognizes. And if you need film or TV dubbing with perfect lip sync, this doesn't solve that part: it's still a voice model, not a facial animation tool. Availability by region and language is also still rolling out gradually.
How to apply it in your business?
If your team already produces content in one language and wants to reach others, cloning the host's voice once (with their consent) lets you generate every new episode in multiple languages from the same translated script, directing emphasis line by line just like with a human voice actor. And if your business wants its automated messages — confirmations, reminders, shipping notices — to sound like the founder's voice instead of a generic one, cloning it once with consent recorded inside the system covers every new message without recording each one separately.
If you already saw the short about this same feature, here's the angle that digs into consent and the real cost: 30 seconds: that's how Gemini 3.8 clones your voice.
Frequently asked questions
How does Gemini verify you have permission to clone a voice?
It asks for a second recording of the voice's owner saying a consent phrase out loud, and only activates cloning if it matches the original sample.
Does Gemini replace ElevenLabs?
Not yet. ElevenLabs is built for studio production with very fine performance tags; Gemini aims for anyone to clone their voice with verified consent without a studio account.
How much does dubbing with Gemini cost versus hiring voice actors?
It isn't free (the API charges per generated character), but the cost moves from hundreds of dollars per language to cents per thousand words.
Does the consent lock stop someone from cloning my voice without permission?
Only within Google's own platform. It doesn't stop another tool, or audio of you that already exists elsewhere, from being used for the same purpose.
Where is Gemini 3.8 Flash TTS available already?
In AI Studio, the Gemini API, Gemini Enterprise, and rolling out to Gemini Notebook and Google Vids too.
Conclusion
The fact that Gemini clones your voice with consent changes the math on who can dub content or give a brand voice to its messages, but the consent lock only solves part of the problem: it protects what happens inside Google's platform, not what happens with audio of you that already circulates elsewhere.