We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Ship realistic
text-to-speech
inside your app.
Wfloat makes speech AI that runs on-device instead of in a data center.
Pay by user count, not usage fees.
Wfloat currently supports these platforms with more on the way.
Web
Browser package setup guide for web apps.
Open Web docs
React Native
iOS and Android package setup guide.
Open React Native docs
Missing the platform or framework you need? Please request it!
Wfloat is for businesses that want to give their AI a voice at scale. It's made for AI powered games, assistants, agents, tutors, and more.
We train and distribute lightweight, lifelike speech models that are small enough run on consumer devices. Rather than chasing frontier model capability, we are zeroing the cost of speech AI at inference time.
In doing so, we will help power this next generation of immersive AI driven apps and products.
Cold water ripples around your boots. Broken banners sway overhead, and something heavy shifts in the dark beyond your lantern.
Left passage. Fresh mud on the stone, and whatever dragged it there is still close.
Then keep your voice down. This hall listens, and the idol below wakes for frightened guests.
Then the surface splits. A brass eye rolls open below you and fixes itself on your lantern flame.
A Mars trip usually takes six to nine months. It mainly depends on the route and where Earth is in its orbit relative to Mars. We can't launch just anytime, because the planets are constantly moving. So missions work best during specific launch windows when the distance and fuel requirements are favorable. That is why mission planners wait for the right alignment, instead of just pointing a rocket at Mars and leaving on any random day.
It's easy, just install the Wfloat package into your frontend app ( takes 5 minutes ).
From there, Wfloat automatically serves the correct speech model assets directly to your end users.
Pricing is different and competitive compared to other speech AI platforms. We don't charge you anything for how much speech audio your users generate. In other words, inference is free.
We charge a small monthly fee for each active user of text-to-speech. For details on cost or how we define an active user, see the pricing page .
It runs on basically any phone, laptop, PC, or tablet from the last decade.
If a user can comfortably run a modern website or mobile app, they will be in good shape here too.