Google Launches Gemini 3.8 Flash TTS With Authorized Voice Replication

Google has launched Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, expanding its generative artificial intelligence (AI) capabilities with new text-to-speech models that can create custom voices and replicate authorized voices using just 30 seconds of reference audio.

The models support more than 100 languages and dialects and are designed for applications ranging from voice agents and podcasts to audiobooks, games, dubbing and other large-scale audio production.

 

TL;DR

 

• Google has launched Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS for AI-generated speech.
• Gemini 3.8 Flash TTS can replicate an authorized voice using a 30-second audio sample.
• The models support more than 100 languages and dialects, along with controls for emotion, pacing and speaking style.
• Google is using consent verification, SynthID watermarking and C2PA credentials for replicated voices.
• Voice replication through Google AI Studio is not currently available in India and several other regions.

 

According to Google's Gemini 3.8 TTS announcement, the company has developed the new models to give creators and developers greater control over how AI-generated voices sound and behave.

Gemini 3.8 Flash TTS is the more capable model, focusing on creative control, voice design and high-quality speech generation.

Gemini 3.8 Flash-Lite TTS, meanwhile, is designed for applications where cost and scale are more important, including dubbing, voice agents and other high-volume speech workloads.

The models are rolling out through the Gemini API and Google AI Studio, while Google is also bringing them to some of its existing products.

 

Gemini 3.8 Flash TTS Can Replicate A Voice From 30 Seconds Of Audio

 

One of the biggest additions is voice replication.

Google says Gemini 3.8 Flash TTS can reproduce the characteristics of a person's voice from approximately 30 seconds of reference audio.

However, Google is placing restrictions around how the capability can be used.

Users must own the voice being replicated or have the necessary rights to use it. Google also says its system includes consent verification designed to confirm that the voice owner has authorized the replication.

Generated content receives additional provenance protections.

Google says audio created using replicated voices includes SynthID, the company's invisible watermark for AI-generated content, as well as C2PA credentials that provide information about how digital content was created or modified.

These safeguards are particularly significant as increasingly realistic AI-generated voices raise concerns around impersonation, fraud and unauthorized voice replication.

As Gadgets 360 reported, the capability is being introduced alongside Google's broader improvements to speech generation and custom voice creation.

 

Google Can Also Create Entirely New AI Voices

 

Users do not necessarily need an existing voice sample.

Gemini 3.8 Flash TTS can also create new voices from natural-language descriptions.

A developer could, for example, describe characteristics such as age, accent, tone, personality or speaking style and have the model generate a corresponding voice.

Google is positioning this capability for creative applications where developers need characters or distinctive voices without relying on recordings from real speakers.

The models also offer more granular control over speech.

Users can provide instructions governing elements such as emotion, pacing, accent and delivery style.

Gemini 3.8 Flash TTS can additionally generate conversational details including laughs, sighs and other non-verbal sounds intended to make synthetic speech feel more natural.

The model also supports native two-speaker generation, allowing developers to create conversations between distinct voices within the same generation.

 

Gemini 3.8 TTS Supports More Than 100 Languages

 

Google says its new TTS models support more than 100 languages and dialects.

That makes the technology potentially useful for global applications such as multilingual dubbing, localized podcasts, audiobooks, customer-service agents and accessibility tools.

Long-form audio generation is another focus.

According to Google's Gemini API documentation, developers can use Gemini's speech-generation capabilities to control characteristics such as style, accent, pace and tone through natural-language prompts.

The company is also positioning Flash-Lite as the more economical option for applications that need to generate substantial amounts of audio.

That distinction could make the model particularly relevant for businesses deploying voice agents or producing multilingual audio content at scale.

Voice Replication Is Not Available Everywhere

 

Despite the broader rollout, one of Gemini 3.8 Flash TTS's most notable capabilities comes with geographic restrictions.

Google says voice replication through Google AI Studio is currently unavailable in India, Illinois and Texas in the US, as well as the European Economic Area, United Kingdom and Switzerland.

The restriction means developers in those locations will not necessarily receive access to every feature highlighted as part of the broader Gemini 3.8 TTS launch.

Other speech-generation capabilities may still be available depending on the product and region.

Google has not indicated when voice replication could become available in the restricted markets.

 

Google Is Expanding Gemini Beyond Text And Images

 

The new TTS models extend Google's broader push to make Gemini a multimodal platform capable of generating and understanding information across different formats.

Voice is becoming an increasingly important part of that strategy as technology companies race to build more natural AI assistants and agents.

High-quality speech generation could allow those systems to interact with users through persistent characters, recognizable authorized voices and more expressive conversations rather than conventional synthetic speech.

Google is also bringing the technology into its own products.

Gemini 3.8 Flash TTS is being introduced in Gemini Notebook, while Flash-Lite is being integrated into Google Vids, according to the company's announcement.

For developers, both models are available through the Gemini API and Google AI Studio.

The bigger question will be how widely developers adopt voice replication and custom voice creation as Google balances increasingly realistic synthetic speech with the safeguards needed to prevent misuse.

With just 30 seconds of authorized reference audio now potentially enough to reproduce a voice, that balance could become increasingly important as AI-generated speech becomes harder to distinguish from recordings of real people.