Built for real conversations
Three people on a call read like a chat, not one unbroken paragraph — SubLive tells the voices apart as they speak.
Know who said what, while they're still saying it
A single running line of captions works fine for one voice. The moment a second person talks — a call, a panel, an interview — it turns into a wall of text with no way to tell who said what. Dialog mode splits it by speaker and labels each turn, live.
Free minutes after you confirm your email. No card required.
One running line, or split by speaker
Subtitle mode
We should ship Friday. I think Friday's tight, let's aim Monday.▍
Dialog mode
Alex: We should ship Friday.
Priya: I think Friday's tight, let's aim Monday.▍
Why it matters
Switch any time
Prefer a single running line? Subtitle mode does that instead — swap between the two from the extension, mid-conversation if you want.
Works wherever the sound plays
Google Meet, Zoom Web, a livestream, a podcast with two hosts — dialog mode does not care what is in the tab, only how many voices are in it.
How it works
- 1
Choose Dialog
In the popup, open Appearance and set Display to Dialog. It is already the default for new accounts. The subtitle window has the same switch, so you can change it in the middle of a conversation.
- 2
Start subtitles on the tab
SubLive separates the voices as the audio plays.
- 3
Read it like a chat
Each turn starts with its speaker and keeps one colour for the whole conversation. You can scroll back, and a button takes you to the latest line.
What it looks like on a real call
Each turn starts with its speaker, so a call with three people reads like a chat. Each voice keeps its colour for the whole call.
Who is who
Speech recognition separates voices by sound; it does not know names. On Google Meet, Microsoft Teams, Zoom in the browser and Jitsi, SubLive watches which participant’s tile lights up while each voice speaks and puts the name on the voice when the match is clear. When it is not clear, the voice stays a number. Names live on your screen only: they are not sent to the server and not saved in the transcript.
An example
In a weekly sync with colleagues in Warsaw, Madrid and Kyiv, a single running line pastes a reply onto the end of someone else’s sentence. In dialog mode the same minute reads as three short turns, each with its speaker, so you can see at a glance that the date was proposed by one person and questioned by another.
What it will not do
- Voices are separated by sound. Two people talking over each other, a bad line or two similar voices can be merged or split, so a label is a guess, not an identification. SubLive does not build voiceprints or match a voice against any other conversation.
- Names appear only where the call platform shows who is speaking. Elsewhere, and in saved files, speakers are numbered.
- Dialog mode changes how lines are shown, not what they cost or which languages are available.
Questions people ask
- Does dialog mode cost more than subtitle mode?
- No. It changes how lines are shown, not how they are billed.
- Can I rename a speaker?
- Not inside SubLive. Names come from the call platform where it can be read. In an exported file speakers are numbered, so replace them in your editor.
- Why does one person sometimes show up as two speakers?
- Voice separation works on sound. If someone changes microphone, moves away from it or the connection degrades, the same person can be split in two. Labels are for orientation.
- Is dialog mode on by default?
- For new accounts, yes. You can change it at any time, in the popup or in the subtitle window.
How dialog mode compares with the speaker names in Google Meet’s own captions is covered in Google Meet translation: built-in captions, speech translation and a Chrome extension.