Using it for the first time? “Three-minute start” is enough to go live. Come back to the section you need when something comes up or you want more out of it.
01
Three-minute start
Oralion has two pages. The speaker's console listens, recognises and translates; the audience's page shows the captions and plays the translation. Viewers install nothing and need no account.
Open the console
Open “Broadcast” at the top of the site in Chrome or Edge on a computer. You can try it without signing in; a free account gets a free allowance every month.
Answer the setup wizard
What language you speak, which languages to translate into, and where the sound comes from. Leave anything you are unsure of at its default; everything can be changed later under the settings.
Press “Go live”
The browser asks for the microphone; allow it. Under the button you can see roughly how much this setup uses per hour.
Share the audience link
Once live, an audience link, room number and QR code appear. People scan or tap it, pick their language and follow along.
02
Before going live
Each time you open the console, the setup wizard walks you through it one question at a time. Once it is closed, the settings card shows the current setup as four stations — Capture → Recognise → Translate → Deliver — with the options below in the same order.
To go through the wizard again, press “🧭 Setup wizard” on the settings card; to stop it opening every time, tick “Don't show this automatically” at its foot. The language button at the top right switches the console to another interface language.
Languages
Pick the main spoken language, then add up to four languages to translate into — five in all. With two languages translation runs both ways: if someone switches to the other language, the captions translate back the other way. More languages cost more per hour.
Choosing “No, original captions only” under “Translate?” turns speech into text and nothing more, at the lowest rate. Target languages show as chips — × removes one, “+ Add language” adds one — with the hourly rate for that many languages beside them.
Where the sound comes from
Microphone
For people speaking in the room. Choose which microphone under “This device”; the clearer the sound, the better the captions.
This computer's sound
For online meetings, webinars and videos. When you go live Chrome asks what to share: choose the tab and tick “Share tab audio” — that is the most reliable choice.
For a desktop app's sound (a meeting app outside the browser, say): on Windows choose “Entire Screen” and turn on “Also share system audio”; on a Mac first allow Chrome under System Settings → Privacy & Security → Screen & System Audio Recording, then restart Chrome.
This device: microphone and echo cancellation
Open “This device” at the bottom of the settings to choose which microphone to use (a USB microphone, a feed from the sound desk) and whether echo cancellation is on. Both are remembered on this computer, so you set them once.
Echo cancellation is on by default: it removes sound that comes out of a speaker and back into the microphone. Keep it on whenever a speaker in the room is playing; if the microphone comes from a mixer, there is no speaker in the room, or the voice sounds thinned out, try turning it off.
Background for this talk
What recognition gets wrong most is names — people, companies, products — and abbreviations. Paste in the topic, the speakers' names, the special terms or the script, and the whole talk comes out noticeably better. Signed in, AI turns it into recognition hints you can check before going live. If you leave it empty, one is worked out from the first five minutes of the talk.
The organised result appears below: “Background” is the topic and field, “Important terms” the words to get right, and “Translation mappings” how a name should be translated (Cirebon ⇄ 井里汶, say). Glance at it before going live; if something is wrong, change the background and let it organise again. To fix how a name is translated, write it into the background.
AI live correction
This option appears once you are signed in.
Normal
Right for most events. AI finds misheard names and terms, fixes the captions already shown, and gets the same word right every time it comes back.
Fast
Corrections arrive sooner — for events where accuracy matters most. Costs a little more per hour.
Full rewrite (paid plans)
Fast correction, plus each passage is polished by a language model once it has been spoken, original and translations alike. The audience sees the sentence rewritten smoothly in place.
Off
Captions never pass through our servers at all; they reach the audience end-to-end encrypted.
Reading the translation aloud
The translation can be read out in a natural voice. Simultaneous reads as the speaker talks, and each listener hears it on their own phone with headphones; consecutive waits for a pause and reads the whole passage, which suits two people taking turns. It is only charged while someone is listening.
Speaking voice
The list is grouped by accent, with voices matching the output language first — they sound most natural; each is marked male or female, and hovering shows a short description.
Where to play interpretation
“Viewers only”: interpretation plays only on each viewer's phone. “Headphone monitor”: the speaker or staff also hear it from this computer — headphones required. “Room speakers (two-way)”: the translation plays over the room's speakers so speakers of two languages can talk in one space; exactly two languages, with this computer's microphone. It tests the speakers and echo cancellation when you go live and filters out feedback automatically.
Consecutive needs two languages
Consecutive works with exactly two languages and pauses listening while it plays; with three or more, use simultaneous so voices do not overlap in the room.
Speaking rate, pause before reading
Adjust these on air under the on-air card: the rate from 0.8× to 1.2×, and for consecutive, how many seconds of pause (1 to 5) before reading starts — longer for a speaker who pauses often.
Other options
Noise reduction
Turn it on for wind or crowd noise; off in a quiet room or with a microphone that already reduces noise. Off by default on phones.
Sentence pause
How long a pause ends a sentence. The default suits most people; pick a longer one for a slow speaker, someone who stops to think, or someone reading a script, so sentences are not cut in half. It can be changed on air.
There are four: Balanced (up to 1.5 s), Conversation (2 s), Slow speech (2.5 s) and Long pauses (3 s). A longer one does not slow the live preview, it only closes each sentence a little later; changing it on air does not interrupt the captions.
Audience questions
Turn it on and the audience's page shows an “Ask a question” button; it can be switched on air too — see “Audience questions”.
Room number and password
You can choose a memorable room number. With a password, only people who know it can watch.
My setups
For a recurring event, save the whole setup — background included — and apply it in one click next time. Up to five, without the room password or this talk's rules. Signed in, it follows you to another computer.
03
While on air
Once live, the settings fold away. From the top: the controls, the audience link, “Adjust on air”, the on-air monitor and live captions.
Adjust on air
Only what is worth changing mid-talk: sentence pause, speaking rate, the consecutive pause, plus rules for this talk, proofreading, audience questions and the correction log. Everything else is locked while live, so nothing changes by accident.
On-air monitor
“Viewers” shows how many people are watching now, and “Last sent” the last caption that went out — a quick check that captions are really reaching people.
Live captions
The language buttons at the top show one language only; “📌 Lock to latest” keeps the newest caption in view, lets go when you scroll back, and “↓ Jump to latest” returns. There is a full-screen button in the corner, and in Chrome or Edge on a computer a floating caption window, so you can keep an eye on captions while presenting slides.
Fix a caption directly
Click any caption in the live caption panel to edit it and press Enter; the audience sees the change at once.
Rules for this talk
When the same word keeps coming out wrong, add a rule (say “Yanbian” → “Yanbian University”). Captions already shown and every later one are fixed, and so are their translations and any speech not yet played.
Correction log
Lists every correction AI made. If one is wrong, delete it here.
Pause and resume
At a break press Pause; the transcript, rules and settings are kept, and Resume carries on. Five minutes of silence also pauses automatically, so nothing is charged for an empty room.
While paused the room stays open, so the audience need not rejoin; only after 30 minutes with no speech does it close. Keep the console page open while paused, or the same session cannot be continued.
End session, start a new one
“End session” saves the transcript and ends the broadcast. “Start a new session” afterwards begins a new transcript with the same settings and background, and can carry this session's rules over too. To split one event into parts, pause rather than end.
Feedback afterwards
When it ends you are asked for a rating and comments; no captions or audio are sent. Signed in, a comment of 30 characters or more earns 0.5 Credits.
Running low
When Credits are running low, a warning and a top-up link appear by the button. Topping up on air carries straight on; there is no need to start again.
Broadcasting from a phone
A phone can only use its microphone. Keep Oralion in front and do not let the screen lock; switching to messages or another app can stop the sound.
04
On a big screen: making the text big
To show captions to the whole room, project the audience's page, not the console:
Open the audience link on the projecting computer
Copy the audience link from the console and open it on the computer connected to the projector — or in a new window on the same computer.
Go full screen
Press F11 on Windows, or Control + Command + F on a Mac. Press it again to leave full screen.
Make the text bigger
Press Ctrl and plus on Windows, or Command and plus on a Mac, a few times until the back row can read it; Ctrl (or Command) and 0 returns to normal.
Show one language
In the settings at the top right of the audience's page, choose the language to project; one language reads more easily. Choose “All” for both.
05
The audience's page
You can send this section straight to your audience. No app to install and no account needed.
Getting in
Scan the QR code on the speaker's screen, tap the link they shared, or open the viewer page and enter the room number. If the room has a password, ask the speaker.
Choosing a language
The first time, you are asked which language to read and whether to hear it. Change it any time from the settings at the top right; Chinese can be shown in Traditional or Simplified.
Text too small
Turn your phone sideways or pinch to zoom; on a computer press Ctrl and plus (Command and plus on a Mac) and F11 for full screen. Showing just one language also helps.
Hearing the translation
Press “Read aloud” at the bottom to hear your chosen language. Please use headphones; if your phone blocks playback, tap it once more.
Back to the latest
Scrolling up to reread stops the captions where you are; press “Jump to latest” to return to what is being said now.
Floating captions (Chrome, Edge on a computer)
Press “Float captions” and they move into a small window that stays above everything, so you can follow along while using other programs. Make the window bigger and the text grows with it; it also has a dark mode.
The screen stays awake
Your phone's screen stays on while you watch, as long as the page is in front.
06
Proofreading together
For an important event, a few colleagues who know the languages can correct captions while it is live, and the audience sees each fix at once. Available on paid plans (Pro, Max, Business), with up to eight proofreaders.
For the speaker
Once live, find Proofreading in the on-air card, press “Create proof link” and copy it to your proofreaders. It is not the audience link — give it only to them. To stop someone, press “Replace link” and the old one stops working.
For proofreaders
Enter your name and languages
Open the link, enter your name, tick the languages you will check and choose a text size. All of it can be changed later under “Languages shown”.
Pick a line, fix it
Click a line to select it, press “✎ Edit” or double-click to change it, and Enter to send. The sentence still being spoken can be edited once it ends; a line someone else is editing shows their name.
Tick what is right
Press “✓ Looks right” on lines you have checked, so the others know. Words with a wavy underline are where recognition or translation was unsure — look there first.
The same mistake keeps appearing
Use “Rule for the whole talk”: replace this… with…, and every caption so far and every later one is fixed. “Undo” takes back a change.
07
Audience questions
Somewhere to ask for people who will not raise a hand. Turn on Audience questions in the setup or the on-air card and a “Ask a question” button appears on the audience's page. People give a name, a little about themselves and the question; everyone sees it, each in their own language.
On the console, questions waiting for an answer are at the top; mark them answered or hide anything inappropriate. They stay with the session and can be downloaded afterwards.
08
Afterwards: transcripts and summaries
Download the transcript
When it ends, download the combined transcript or a single language from Transcript archive on the console. By default it is kept in this computer's browser.
Keep it in your account
Turn on “Save my transcripts to my account” in your account settings, and every session is saved there and readable on any computer.
The Transcripts page: your meeting library
Signed in, press “Transcripts” at the top. Live sessions and transcribed recordings are in one list, by date, one line each: time, title, who took part, languages, and where it is kept.
Summary and full text
Click a line for its AI summary: who took part, what was discussed, decisions and next steps. “Open” shows the full text on the right (above, on a phone), so you can read one session while finding the next. A live session switches between its languages; a recording between the AI-improved and raw versions.
Edit a summary
If a summary is wrong or missing something, press “Edit” and change the title and points directly, one per line.
Search and filter
The search box finds titles, names, file names and even things that were said; for words in older transcripts saved to your account, press “Search account text”. A date range sits below it. Click any participant's name to list only sessions with that person — “what did I discuss with them?” in one click.
Categories
Like folders, each session goes in one category. “+ New category” creates one; file a session from the Category menu when it is open. “Auto-file” files the unfiled sessions by their summaries and never moves what you already filed.
This device, account, or both
Each line shows where it is kept. A session only on this device cannot be seen from another computer; with “Save my transcripts to my account” on, “Sync this device” under the search box uploads them.
Timestamps, download, delete
The “⋯” in the reader shows the time of each passage; download and delete are at the foot of an open session. Once deleted from your account, other devices will not sync it back.
09
Transcribe a recording
For a meeting, interview, lecture video or field recording you already have: sign in, upload it under “Transcribe”, and come back when it is done — no need to watch it.
Choose the file
MP3, M4A, WAV, MP4 and other audio or video files, up to 200 MB each.
Choose the settings
Leave the main language on auto-detect if unsure; turn on “Identify speakers” for several voices; for the AI pass, “Standard” fixes recognition errors, punctuation and paragraphs, while “Aggressive” also removes fillers like “um” and “you know” so it reads more smoothly.
Start
After “Start transcription →” the file is uploaded, recognised and then tidied by AI. Progress shows under recent transcriptions; you can cancel while it runs and get the Credits back, and a failure refunds them automatically.
Read, compare, export
When finished, switch between the AI-improved and raw versions; “Track changes” colours what AI removed and rewrote. The raw version can show the time of each passage. Copy everything, download a TXT, and find it on the Transcripts page with a summary.
10
Account, plan and billing
Account settings
Overview
At the top: your plan, how much you have used this month and what is left, and your permanent Credits.
Keeping transcripts
By default captions travel only between your browser and your audience's, not through our servers. Turn on “Save my transcripts to my account” and finished sessions are saved there; the most recent are listed below to download or delete, or delete them all at once.
Language for summaries and titles
Whatever language a meeting was held in, its AI summary is written in the language chosen here, so the whole list reads at a glance. Sixty languages to choose from.
Signing in
Change your email or set a password; you can also sign in with Google or with a link sent to your inbox. To stop product news, untick “Receive product news and updates by email” — sign-in links and receipts still arrive.
Billing and subscription
Press “Billing →” in your account settings to see your plan, next charge date, this month's allowance and the weekly Voice allowance; upgrade, buy permanent Credits and see your payment history there. To cancel, change the card or download invoices, use “Manage · invoices”.
Usage draws on the plan's allowance first and only then on permanent Credits, which never expire.
11
Settings for your event
Talks, classes, lectures
Sound: microphone, close to the speaker or fed from the sound desk.
Background: title, speakers' names, special terms — paste the whole script if there is one.
Reading a script or speaking slowly: a longer sentence pause.
Project the audience's page from another computer, full screen.
Online meetings, webinars, videos
Sound: this computer's sound, in Chrome or Edge.
When sharing, choose the meeting's tab and tick “Share tab audio”.
Paste the audience link into the meeting chat.
To follow along yourself, the audience page's floating captions sit over the meeting window.
Two-language conversations, interviews, meetings
Choose exactly two languages; translation runs both ways automatically.
Sentence pause: “Conversation”.
To hear the translation: “consecutive” for taking turns; in one room without headphones, simultaneous with the room speaker option (microphone only).
International events, mixed audiences
Add 3–5 languages; each person chooses their own.
For speech, simultaneous — each listener on their own phone and headphones.
More languages cost more; check the estimate under the button before going live.
For the highest accuracy, use fast correction or full rewrite, with proofreaders.
Tours, outdoors, on the move
You can broadcast from a phone; turn off auto-lock and keep Oralion in front.
Turn on noise reduction in wind or crowds; a clip-on microphone helps a lot.
Listeners use headphones for the read-aloud and do not have to watch the screen while walking.
Internal or confidential meetings
Give the room a password.
Set AI correction to Off and captions never pass through our servers.
Transcripts stay on this computer unless you turn on keeping them in your account.
Weekly meetings, long events
Save the setup under “My setups” and apply it in one click each time.
Check the time shown under the button: will it last the whole event?
Pause during breaks; nothing is charged while paused.
12
Credits and cost
Usage is counted in Credits, by time actually on air; nothing while paused. More languages, read-aloud, fast correction or full rewrite use more per hour.
Before you go live, the line under the button says what this setup uses per hour and roughly how long your balance lasts. If it may not last the event, it tells you to top up first.
A free account's allowance refills every month; monthly plans include plenty of hours; and Credits you buy outright never expire.
Watch the level meter at the top of the console. If it does not move, check the browser is allowed to use the microphone, pick another one under “This device”, or check the computer's input volume.
No captions when sharing computer sound
Choose a tab and tick “Share tab audio”; a whole screen or window only carries sound if a “Share system audio” option appears. Use Chrome or Edge.
The captions keep repeating themselves
Almost always the translated speech coming out of a speaker and back into the microphone. Whoever is listening should use headphones; the console also shows a warning with a one-click fix.
Names and terms keep coming out wrong
Write those words into the background before going live; on air, add a rule for this talk and every caption so far and to come is fixed.
The audience says the text is too small
Point them to “The audience's page”: phone sideways or pinch to zoom, Ctrl and plus on a computer. For a projector, use the audience's page in full screen.
Listeners hear no read-aloud
Phones often block playback until tapped; ask them to tap “Read aloud” again and check the phone is not muted. Read-aloud must also be on at the speaker's end.
Captions stopped while broadcasting from a phone
A phone stops listening when you switch apps or it locks. Go back to Oralion, press Resume, and turn off auto-lock.
Can someone help me set it up?
Of course — write to [email protected]. Setup advice for non-commercial events, and general questions, are free.