Choose the right session
Use Transcribe only when you need an original-language transcript. If you also want a summary, choose a supported session with transcript capture and summaries enabled.
SalamAI turns speech into live captions and saved text for meetings, talks and events. Keep the original language or translate it. Summaries are available on supported sessions, but not in standalone Transcribe only mode.
Explore in the consoleIn short
Updated
Live captioning shows speech as text while someone talks; a transcript is the saved text you review later. SalamAI does both. Captions can stay in the original language or be translated, and you can show them full-screen on a projector, in a simple text view or on audience phones. Where enabled, Transcribe only mode captures speech in its original language without audience streams, audio recordings or summaries. When saving is on, supported sessions save transcripts as Markdown and captions as SRT subtitle files. Summary-enabled sessions can also produce a structured summary in up to five languages. Check important names and figures, because AI captions can contain mistakes.
Open a caption window for each language and put it full-screen on a projector or display, or use the simple text view. Adjust the text size so people at the back can read it, and choose Words or Sentences display, whichever your audience finds easier to follow. Place the screen where people can see both it and the speaker. For livestreams, send captions to an OBS overlay or a WebVTT feed where enabled. Remember that shared captions need readable text on the sharing service, unlike saved archives, which are encrypted in the browser, so control access before you enable them.
Read nextSupported translation languages →Translation privacy & security →
Meeting transcription
A team discussion, conference talk or learning session: keep a readable record, then return to the ideas that matter.
Use Transcribe only when you need an original-language transcript. If you also want a summary, choose a supported session with transcript capture and summaries enabled.
Connect your microphone or supported shared audio, select the recognition language and enable saving before you start. Follow live text and export Markdown or SRT files after the session.
For supported sessions, request a structured summary in up to five languages. Review the key points, decisions and action items, then copy or download the result for your team.
Caption and transcription options
Turn speech into text in another language without generating a spoken translation. Share supported text-only audience channels or an overlay where enabled.
Show speech as text in its original language, with no translated voice. Display it in the room or share a text-only caption channel.
Capture an original-language transcript in a dedicated session, where enabled. Translation channels stay saved but do not run; this mode does not create audio recordings, audience streams or summaries.
Designed to be read
Welcome to the staff training session.
Today we will review the new safety process.
Please send your questions before the break.
Words shows text as it arrives, so early words may change. Sentences waits for a complete sentence or a fallback timeout. Only confirmed text is saved or shared through caption outputs.
Show up to three caption panes inside the console, or open independent language windows. Use full-screen, a simplified CC view or picture-in-picture where supported. Native captions are available where enabled.
Adjust text size, colours and the reading area, show available original text and scroll back to review. Following pauses while you read earlier lines; return to the latest text when you are ready.
Names and specialist terms
Paste or import names and preferred spellings, then save a draft. Supported One-to-Many captions can match romanised and similar-looking forms; optional AI-assisted checks help review uncertain names where enabled.
The filler filter can remove filler words from supported One-to-Many text and saved captions. Name and filler tools change text only. They do not correct spoken translation or guarantee accuracy.
In Team → Vocabulary, add names, brands and specialist terms you want the system to recognise. Up to 100 terms can guide same-language captions and Transcribe only when recognition hints are enabled.
Use the translation glossary for preferred translated terms. Use Vocabulary to help recognise speech. Vocabulary changes apply when you next press Start. They can help, but do not guarantee a spelling.
Saved text and live feeds
Enable subtitle files, readable transcripts or both. Saves use confirmed text, preserve available language and track information, and encrypt content before upload. Long captures may create multiple numbered files.
Where enabled, send text to an OBS overlay or an external WebVTT feed. These are explicit sharing paths with their own access controls; standalone Transcribe only does not publish them.
Common questions
Yes. Where enabled, Transcribe only captures speech in its original language. Choose the recognition language and enable saving before you start. This mode does not create audience streams, audio recordings or summaries.
Captions show speech as text during a session. A transcript is a saved text record for review later. SalamAI can save supported transcripts as Markdown and captions as SRT subtitle files when saving is enabled.
Supported sessions can create an optional structured summary in up to five languages. Standalone Transcribe only does not create summaries. Choose a summary-enabled session before starting if you need both. See summary options.
Plan your event
Tell us your languages, crowd size and audio setup. We’ll show you SalamAI live and suggest the right setup for your event.
Already have an account? Open the console →