The suit is perfect.
Now comes the voice.
You spent months bringing her to life. One sentence and the illusion drops. So you go quiet, or you haul a laptop rig in a backpack. Vellic is the third option: a device that lives in your suit and speaks in her voice, live, as you talk.
One take, no edits, no PC.
This is me testing the prototype in my bedroom, unscripted, straight off my phone. The device on the lanyard is converting my voice live as I talk, on its own onboard compute, nothing else in the loop. It's rough on purpose: this is exactly what the hardware sounds like today, not a produced demo. Cleaner footage and board photos will land here as the build tightens up.
What it actually is
Not an app, not a cloud service, not a soundboard. Vellic is dedicated hardware built around one job: convert your voice fast enough to hold a face-to-face conversation, all day, inside a suit.
Streaming conversion, mouth to speaker in a fraction of a second. Banter-speed, not press-and-hold.
Onboard battery sized for hours of floor time. No outlet hunting between photo ops.
No PC, no phone in the loop, no internet. Your voice never leaves the device.
Physical controls you can find through a suit, and in-ear monitoring so you hear her voice, not yours.
Her voice, not a preset
Target voices are trained per character. You're not picking from a dropdown of ten stock voices. Voice training will be offered alongside the device, priced separately (details TBD), and we're exploring letting you train your own models if you'd rather keep it entirely in your hands.
And to be clear: any target voice works, masc, fem, or neither. We're starting voice-first with fem fursonas because that's where the gap between performer and character is most often the biggest, but if your character needs a voice, this is for you too.
What it'll cost
That's the honest number, and here's why: the AI compute module inside is several hundred dollars of silicon on its own, before the board, battery, audio hardware, and enclosure around it. This is fursuit-tier pricing for a fursuit-tier piece of kit. Final pricing may move as the production build gets locked, and the waitlist hears first if it does.
Who's building this
One engineer, in the open. I build real-time audio systems. The conversion engine is written in Rust and runs on embedded NVIDIA hardware. My voice is the main reason I don't suit myself. Once I had a fix running on my desk, I realized how many other people are stuck on the same thing.
No pre-orders are open and no money is being taken. Right now I'm finding out how many people want this to exist. That's what the list below is.
Get on the list
Waitlist gets new demos first and pre-order access first, with pre-orders targeted for 2027, and a total of maybe five emails ever. No name needed. This list stays private and dies with the project if the project dies.