bubbl!
See every word.
Real-time captions that float near each speaker's face, so you always know who said what.
Launching soon on
- Android Coming soon
- iOS Coming soon
- AR glasses Coming soon
Captions belong near the speaker.
Most captioning apps put text at the bottom of the screen. You read the subtitle, then look back up to see the person's face, then look down again when they start speaking. In group conversations you lose track of who said what entirely. bubbl! places each caption near the face it came from, so you can read and watch at the same time.
Text at the bottom. Speaker unclear. You look down, then back up, then down again.
Each bubble stays near its speaker. No looking away. No confusion.
Built for real conversations.
Speaker-anchored bubbles
Captions attach to the face they came from. In a three-person conversation, three separate bubbles track three separate faces.
Multi-speaker diarization
Automatically tells speakers apart using audio and face identity, so captions stay sorted even when multiple people talk close together.
Lip landmark anchoring
Bubbles snap to the mouth, not just the face bounding box. As you move around, the caption moves with you at 30 frames per second.
Emotion and tone
A small tone label next to each caption, joyful, frustrated, warm, or confused, restores the feeling behind the words that plain text hides.
Directional audio cues
When a speaker is off screen, an arrow points toward them. Stereo mic analysis tells you left, center, or right before you see the face.
Conversation summaries
After a conversation ends, a short AI-generated summary appears in history. Key points, action items, and who said what, in one line.
AR glasses mode
On compatible AR glasses, captions float in your field of view hands-free. No phone to hold. No screen to look down at.
Cardboard VR fallback
No AR headset nearby? A Google Cardboard viewer turns any Android phone into a stereoscopic captioning visor for under ten dollars.
Accessibility-first decisions
Adjustable font size, bubble opacity, and contrast. Translation into English from 44 languages. Every setting is accessible without opening menus.
Everyday conversations, made legible.
-
Doctor visit
Follow every word of a medical consultation.
No interpreter needed. Point the phone at the doctor and read their words as they speak.
-
Family dinner
Keep up when six people are talking at once.
Each voice gets its own bubble. Crosstalk stays sorted because each bubble is tied to a face.
-
Classroom
Hear the lecture, then review the summary.
The transcript saves automatically. After class, the AI summary becomes your notes.
-
Job interview
Focus on the conversation, not on lip-reading.
Each panel member gets their own bubble. Afterwards you can review who asked what.
Three surfaces, one experience.
-
Android
Coming soonWorks on any Android phone with a camera. No special hardware. Open source, available on Google Play when it launches.
The current Android build is already running. Public release coming soon.
-
iOS
Coming soonThe same face-anchored captions, the same conversation history, on iPhone and iPad via the App Store.
iOS launch follows the Android release. Join the waitlist to hear first.
-
AR glasses
Coming soonCaptions in your field of view, hands-free and socially invisible. No phone to hold. Compatible with Snap Spectacles, Xreal, and other waveguide glasses.
This is where deaf users actually need this most. It's the next step after the phone release.
Be first to know when we launch.
We will send one email when bubbl! is ready on Android, iOS, or AR glasses. No spam, no newsletter, just a single launch notification.
Built by a small team for a large need.
bubbl! started as an entry to TSA Software Development 2026, where we competed at the national level. The problem we set out to solve, captions that follow the speaker instead of sitting at the bottom of a screen, turned out to be one nobody had fully solved yet. So we kept building.
The codebase is open source on GitHub. Everything from the face-anchoring layout algorithm to the speaker diarization pipeline is public. We believe accessible technology should be inspectable, improvable, and free.
