Transcribe a live stream after it ends: keep the recording, open it from Past streams, run automatic subtitles, fix the text and download TXT or SRT.
Updated · 4 min read
StreamLive Pro transcribes a live stream from its recording, after the broadcast ends. It does not show captions to viewers during the live. Press Record during the broadcast so the recording is kept, then open it from Past streams, run automatic subtitles, correct the text and download it as a plain transcript (TXT) or as a caption file (SRT or VTT).
Before you start
Press Record in the studio. A temporary copy of the broadcast exists for the preview and for cutting Shorts, but the recording is kept only if you pressed Record. Otherwise it is removed about 2 hours after the broadcast. Details: Recordings and storage — what is kept and for how long.
Plan for sound, not only picture. The transcript is as good as the audio. A microphone close to each speaker matters more than the camera. Several people talking at once lowers accuracy the most.
Minutes. Transcription runs in Translate and subtitles; how many minutes each plan includes is on the /pricing?lang=en. The 7-day trial handles videos of up to 10 minutes there, so a full-length broadcast needs a paid plan.
End the broadcast and wait a moment. The recording is processed after the live ends, usually within 1 to 3 minutes. If you open it too early, the system tells you and opens it by itself when it is ready.
Open Past streams. In the menu choose Past streams and find the broadcast.
Open it in the editor. Choose Edit Video on the broadcast. The recording is copied into your library and opens in the editor.
Run automatic subtitles. Open Subtitles and choose Auto Subtitles. Speech becomes text with a time for every word, split into lines at natural pauses.
Correct the text. Click a line to jump to it and fix names, terms and numbers. For a long broadcast, fix the recurring names first with find and replace.
Download the transcript. Press Export. In the Subtitles tab, TXT gives the text without times, for an article or show notes. SRT or VTT gives timed captions for YouTube or an editor.
If you streamed from OBS or vMix and recorded on your own computer, or the cloud recording was already removed, upload your local file: Translate and subtitles > Upload new video for translation, with the same language as both source and target. The rest of the steps are the same.
What you get, and what you do not
Output
What it is good for
TXT
Plain text of everything said: a summary, a blog post, a searchable archive of classes
SRT / VTT
Captions with times: YouTube captions for the recorded video, Premiere Pro or DaVinci Resolve
Burned-in MP4
The recording with captions in the picture
Shorts
Short vertical clips with captions, made from the broadcast automatically unless you turned automatic Shorts off
What it does not give you: text during the live, dubbing, or a transcript of a broadcast whose recording was removed and that you did not keep yourself.
What can go wrong
No recording on the broadcast. Record was not pressed and more than about 2 hours passed. Upload a local copy if you have one; next time press Record.
The editor says the recording is still processing. Wait; it opens by itself when ready.
Names and terms are wrong all through the text. Speech recognition writes the nearest common word for rare names and specialist terms, for example Aramaic terms in a Gemara class. Fix each with find and replace.
Two speakers merged into one line. Split the line at the change of speaker.
Long silences or music at the start. Trim the recording in the editor first, so the transcript and the captions start with the speech.