這種小型蠑螈是來自墨西哥城的美西螈
OWN LANEFEATURES · SYSTEM OVERVIEW
Six capabilities. One path to understanding.
From capture and live refinement to multilingual captions, interpretation, transcripts, and summaries: this is not a feature checklist. It is how Oralion carries a conversation all the way to understanding.
Follow the path ↓- 60
- available languages
- 1–5
- languages in one room
- E2EE
- captions and interpreted voice
01 / LIVE REFINE
It does not wait until the end to edit. It learns what was misheard while you are live.
Recognition often fails on names, places, and terms that still look plausible. Oralion uses context to find mistakes likely to recur and creates a safe, reusable correction rule.
- Corrects recognition errors without rewriting the speaker's voice or grammar
- Repairs captions that are already on screen
- Applies the right term first every time it appears again
CONTEXT FOR THIS SESSION
A museum talk about unusual amphibians
00:14 · ASR
The tiny salamander is an axle lot from Mexico City
RULE LEARNED
axle lot→axolotl
- PASTCaptions already shown✓
- NOWThe sentence being read✓
- NEXTEvery recurrence ahead✓
ONE VOICE
この小さなサンショウウオはメキシコシティ原産のウーパールーパーです
OWN LANELoài kỳ giông nhỏ này là axolotl đến từ Thành phố Mexico
OWN LANESalamander kecil ini adalah axolotl dari Kota Meksiko
OWN LANE02 / TRUE MULTILINGUAL
One voice does not become five copies of the same translation.
Each target language has its own recognition, translation, and voice path. One room can run one to five languages at once; every listener chooses and receives only the lane they need.
01Recognition context prepared per language
02Captions and voice routed by language
03Every listener switches independently
03 / VOICE MODE
Read when you want. Let the translation speak when you need it.
Translation does not have to stay on screen. With headphones, Oralion reads the chosen language aloud in segments; captions arrive first while voice follows on its own timeline.
SIMULTANEOUS
Simultaneous mode
Starts speaking as each translated segment completes—suited to talks, classes, and videos.
CONSECUTIVE
Consecutive mode
Waits for the speaker to pause, then plays the full translation—suited to turn-by-turn conversations.
04 / SIGNAL PATH
The real difference lives in the path the sound takes.
From capture to the listener's phone, every stage does only the job it needs to do. Speed, accuracy, and privacy do not have to be traded against one another.
-
01
Capture
Microphone, browser tab, or computer audio
-
02
On-device denoise
Reduce steady noise and echo on the speaker's device
-
03
Recognize live
Recognition primed with this session's topic and terms
-
04
Learn corrections
Optional AI correction learns reusable rules
-
05
Branch by language
Translate and create interpreted voice per language
-
06
Encrypt and deliver
Route ciphertext to the right device by listener choice
PRIVACY BY ARCHITECTURE
Oralion's server does not need to read a caption to deliver it to the room.
- Raw audio goes directly from the speaker's browser to the speech provider, not through Oralion's relay.
- Captions and translated voice are encrypted in the speaker's browser; the relay sees only ciphertext and delivery metadata.
- When live AI correction is enabled, the required caption excerpts are an explicit speaker-controlled exception.
Sea otters wrap themselves in kelp when they rest so they do not drift away.
AI SUMMARYKelp works like an anchor for resting sea otters.
05 / AFTER THE MOMENT
Afterward, the live moment becomes something you can return to.
When you choose to keep it, live captions become a transcript; AI then organizes a title, topics, summary, and actions. Recorded classes, interviews, and videos can be uploaded into the same workflow.
- Live sessions and uploaded recordings in one personal library
- Owner-editable transcripts and AI summaries
- Storage is your choice, with durable deletion
THE DETAILS THAT MATTER
Everything you need on air, without demanding your attention.
Three audio sources
A microphone, a browser tab, or meeting and video audio from the computer.
Room code, link, and password
Fast to share publicly, with a password gate when the room needs privacy.
Speaker labels
When people take turns, the transcript keeps track of who said what.
Nothing for viewers to install
Scan a QR code or open a link, then choose a language without an account.
Picture-in-Picture and wake lock
Keep captions above other content and keep the screen awake while following along.
Rooms stay isolated
Each room has its own process, so one busy room does not clog captions for everyone else.
UNDERSTAND IT YOURSELF, THEN SHARE WHEN YOU WANT
Open a room and hear the difference.
No event to organize and no email required before you try it.