On a call? Let your laptop transcribe and translate
What is SubKun Live Transcribe?
SubKun Live Transcribe is a free live transcription tool that runs in your browser. It listens to spoken English, from your microphone or from any browser tab, and turns it into live captions about one second after the words are said. It can show a translation under each caption, and it keeps the full transcript beside them so you can scroll back, copy it or export it.
There is nothing to install and no account to create. The speech-to-text model is downloaded once and then runs on your own computer, so your audio is never uploaded.
Real-time transcription with about one second of delay
Live captions only help if they keep up with the speaker. When the text arrives several seconds late, you are still reading the last sentence while the next one is being said, and in a two-way conversation you cannot answer in time. That delay is why live transcription so often feels unreliable in real use.
SubKun Live Transcribe runs IBM Granite Speech TurboCTC, an open speech recognition model that IBM released in August 2026. At release, IBM reported it as the fastest model on the Hugging Face Open ASR Leaderboard, about twice as fast as the previous leader. It is light enough to run on your own graphics chip, so words appear about one second after they are spoken. While it listens, the page shows the delay it is measuring, so you can check the number yourself.
If English is not your first language, that speed changes what captions are for. You can read the word you did not catch while the sentence is still going, glance at the translation, and reply without asking anyone to repeat.
How to live transcribe audio in three steps
- Open this page in Chrome or Edge on a computer. The first visit downloads the speech model, about 550 MB. After that it loads from your browser cache.
- Choose what to listen to. "Start with microphone" transcribes whatever your mic hears. "Use tab audio" transcribes the sound of a browser tab: pick the tab and turn on "Also share tab audio".
- Read the live captions as people speak. Turn on "Show translation" to add a second line in your language. When you are done, open Export to copy or download the transcript.
What people use live captions for
Meetings and calls
Keep up with a Zoom, Google Meet or Teams call held in English. If the meeting runs in a browser tab, capture that tab. Otherwise let the microphone listen to your speakers.
Live streams and videos
Get live subtitles for a keynote, a news broadcast, a podcast or a YouTube video that has no captions of its own. Tab audio is captured directly, so it works with headphones on.
Lectures and online classes
Read along while the lecturer talks. Afterwards, export the transcript as notes and ask an AI to summarize it.
Phone calls
Put your phone on speaker next to a laptop and read what the other person is saying, with a translation if you need one.
When listening is hard
Real-time captioning helps when the sound is poor, the accent is unfamiliar, or you are deaf or hard of hearing. The captions are large and each sentence stays where it first appeared, so your eyes do not have to chase the text.
Learning English
See every word as it is spoken with the translation underneath, then review the bilingual transcript afterwards.
Live caption translation in 8 languages
Turn on "Show translation" and each caption is translated as it appears, into Simplified Chinese, Traditional Chinese, Japanese, Korean, Spanish, French, Portuguese or Arabic. The English stays on top with the translation below it, so you can check one against the other. When you export, choose English only, the translation only, or both.
Speech recognition is English only for now. Translation is done by an online translation service, so when it is turned on the caption text is sent out to be translated. The audio never is.
Live Transcribe for PC and Mac
Google's Live Transcribe app is only available on Android. SubKun Live Transcribe is a separate, independent tool that does the same job on a computer. It works in desktop Chrome and Edge on Windows and macOS, with nothing to download from an app store. It is not affiliated with Google.
Phones are not supported yet, because the speech model is too heavy for them. During a phone call, put the phone on speaker next to your laptop instead.
Why it runs on your computer
Most live transcription services send your audio to a server and limit how many free minutes you get. Here the model runs locally with WebGPU. That is why captions appear in about a second, and why there is no sign-up, no minute counter and no recording kept anywhere. The transcript exists only in this page, so export it before you close the tab.
Frequently asked questions
Is SubKun Live Transcribe free?
Yes. It is free for now, with no sign-up and no time limit on a session.
How fast is SubKun Live Transcribe?
Text appears about one second after the words are spoken. The exact delay depends on your computer, and it is shown on the page while it listens. Without WebGPU it is noticeably slower, so use desktop Chrome or Edge.
Which speech recognition model does it use?
IBM Granite Speech TurboCTC, a 470-million-parameter English speech recognition model that IBM released in August 2026. It runs entirely in your browser with WebGPU.
How do I transcribe computer audio, such as a video or a live stream?
Choose "Tab audio" as the input, pick the tab that is playing the sound, and turn on "Also share tab audio". The audio goes straight from the tab to the transcriber, so it works even with headphones. To transcribe a desktop app, let the microphone listen to your speakers.
Does it work with Zoom, Google Meet and Microsoft Teams?
Yes, alongside any of them. For a meeting in a browser tab, capture that tab to transcribe what the other participants say. For the desktop apps, use the microphone. No bot joins the call and nothing is recorded.
Is my audio recorded or uploaded?
No. The speech recognition model runs inside your browser, and audio is processed in memory and then discarded. Only when translation is turned on is the caption text sent to a translation service.
Why does the first load take a while?
The speech model is about 550 MB and is downloaded on your first visit. Your browser caches it, so later visits start much faster.
Can I save the transcript?
Yes. Open Export and choose English only, the translation only, or both. Each one can be copied to your clipboard or downloaded as a Markdown file. A common next step is to paste it into an AI assistant and ask for a summary.
Is this an alternative to Web Captioner?
For English captions, yes. Web Captioner, the free browser captioning tool, shut down in October 2023. SubKun Live Transcribe covers the same basic need, free live captions in a browser tab, and adds translation and transcript export.