Google Gemini prepares to let users fine-tune voice models with custom controls

Gemini gets ready to let you fine-tune its voice models

Options are set to include Speed, Energy, Warmth, and Formality. Google Gemini is already an incredibly versatile tool, and that versatility doesn't just extend to the questions it's capable of answering, nor the tasks in can accomplish. We're also talking about just how many options Google gives you for how you'd like to interact with the AI agent, letting you choose between some very distinct voice models: Orbit, Pegasus, Ursa, and more. And while that's already very nice, it looks like Google's currently working to give us a whole lot more fine-grained control over just how these voices sound. Right now, Gemini's voices are largely all-or-nothing affairs: You choose one, and you just get to use it as-is. But checking out the changes present in version 17.41.12.sa.arm64 of the Google app for Android, we've identified work on a system that should eventually let you dial in a number of new parameters for further customizing Gemini's voices. Rather than using the voices as Google has already configured them, Gemini is getting ready to let us tweak four output options: Speed, Energy, Warmth, and Formality. Once these are live, you'll find them as adjustable sliders within Gemini voice settings: Now, as you can probably tell from that message at the bottom, none of this is working at the moment. We've been able to pull up an early preview of the sliders themselves (and you won't even see that much on your own device just yet), but they're not yet wired up to have any impact over the voice output itself. When Google finishes up its work, though, Gemini users will be able to independently make those adjustments to their heart's content. Even with just these basic labels, it's not too difficult to predict what kind of effect each option will have. The fact that we're seeing Speed and Energy as two distinct options makes it likely that the former will be largely just about words per minute, while the latter should let you tweak intensity independently of the output rate. Warmth will presumably control just how friendly and empathetic Gemini sounds, while Formality could involve word choice and just how detailed the responses are. Shades of this already exist, like how we can ask Gemini Live to speed up or slow down, but we don't yet have any sort of settings for making that permanent, nor the ability to work with the rest of these options. Finally, we've spotted this fun little UI behavior Google is cooking up, with the individual Gemini voice names fading in and out as you swipe through the list. It feels likely we'll see both these changes roll out at the same time, as we don't currently have any sort of swipe-like interface for Gemini name selection.

Gemini could finally let you choose how friendly it sounds

Finally, you can stop Gemini from sounding like your HR manager Google has spent the past few months making Gemini sound more natural, expressive, and conversational. Now, it appears the company is preparing to give users far more control over how the AI speaks. Code spotted by Android Authority's APK Insights - the latest beta version of the Google app suggests Gemini may soon allow users to customise its voice across four separate parameters: Energy, Formality, Warmth, and Speed. Instead of choosing from a fixed list of personalities, users could tweak these characteristics to create a voice that better suits their preferences. Recommended Videos The feature hasn't been announced yet, but the discovery points to Google's next step in making AI assistants feel more personal. Gemini's voice is getting more flexible The upcoming controls were discovered in Google app 17.41.12 beta, where new strings reference a dedicated "Customize" section within Gemini's voice settings. According to the code, users will be able to adjust four characteristics: * Energy: Low, Medium, High * Formality: Low, Medium, High * Warmth: Low, Medium, High * Speed: Slow, Normal, Fast Rather than replacing Gemini's existing voices, these settings appear designed to build on them. Once configured, the customised voice is expected to carry over to both Gemini Live and the standard chat experience, ensuring a consistent personality across the app. The approach marks a notable shift. Until now, selecting a Gemini voice has largely meant choosing from Google's predefined personalities. These new controls suggest Google wants users to shape the assistant's tone instead of simply picking one that feels closest. Google already refreshed Gemini's voices after I/O The discovery comes shortly after Google rolled out a refreshed voice selection experience following Google I/O. The redesigned picker replaces the previous carousel interface with a cleaner list view and introduces two new voices: Flare and Glow, replacing the older Nova and Lyra options. The current lineup now includes: * Ursa * Vega * Pegasus * Dipper * Eclipse * Capella * Orbit * Orion * Flare * Glow Interestingly, Google has also removed the descriptive labels that previously explained each voice's personality, such as "Calm" or "Bright." Users are now left to judge the voices by listening to them rather than relying on written descriptions. The update also refreshes Gemini's interface with thinner, more modern icons for features like the microphone, camera, gallery, file uploads, video, screen sharing, and Gemini Live. These visual changes are rolling out through a server-side update alongside Gemini version 1.0.913571982. The timing is also interesting. During Google I/O, the company confirmed that regional dialects are on the roadmap for Gemini. Voice customisation would fit neatly into that broader effort to make AI conversations feel less robotic and more tailored to individual users. Google also isn't alone in moving in this direction. Apple's iOS 27 introduces similar controls for Siri AI, allowing users to adjust Pace and Expressivity, with those preferences extending across Siri-powered experiences like Maps and Safari. Giving users control over tone, warmth, and speaking style may seem like a small change, but it reflects a bigger shift in the AI race. Companies are no longer competing solely on what their assistants can do. Increasingly, they're competing on how those assistants sound while doing it.