ElevenLabs turns X profiles into custom AI voices

ElevenLabs just open-sourced X to Voice, a fun new tool that reads any X (formerly Twitter) profile and automatically designs a matching AI voice based on how that person posts.

What it does

You type in any public X username, and the system automatically inspects the account bio and recent posts to get a read on their personality and style.

From that analysis, it crafts a custom text prompt detailing acoustic traits like pitch, timbre, speaking pace, and intonation, then routes that prompt straight into ElevenLabs' Voice Design model.

In hands-on testing by Gadgets 360, the whole pipeline takes roughly 30 seconds to a minute. It returns three unique voice samples reciting a line customized to that profile, along with the generated text prompt.

If the profile picture happens to be a clear face close-up, the tool goes a step further: it animates the avatar and syncs its lip and mouth movements directly to the spoken audio.

How it works and what developers get

X to Voice is accessible via a web client, and the underlying project is fully open-source.

Alongside this showcase, ElevenLabs rolled out an official API for its Voice Design feature. It comes with two main endpoints: one that generates three distinct voice previews from any descriptive prompt, and a second to save your favorite picks to a library for local use.

That means developers do not have to rely on social media data; they can programmatically generate custom voices on the fly for AI agents, game NPCs, or interactive assistants.

One catch: ElevenLabs hasn't yet shared pricing, per-request costs, or technical architecture details for the underlying model.

Why it matters

Custom voices usually require one of two painful steps: collecting clean audio recordings for voice cloning, or tinkering with confusing audio parameters by hand. X to Voice skips both by turning public social writing into an acoustic personality blueprint.

For creators and indie developers building character-driven apps or virtual agents, this is an immediately useful reference design. If the API pricing ends up being accessible, expect this automated persona-to-voice pipeline to show up across plenty of interactive chatbots soon.