TikTok Live Archive — Recorded Live, Not Fetched Later
An archive can only hold what was recorded while it aired. Nothing on this site reaches back to a broadcast that has already ended.
Recording is queued work
That is the hard constraint behind every live archive: a stream is not a file until someone captures it, so building a collection means starting a job each time an account goes live rather than searching for past broadcasts afterwards. Savett's part is the capture and the handover. The finished file expires about an hour after the job completes, so the archive itself lives in your own storage and not here, and nothing is kept on your behalf. In practice that means collecting each recording promptly and naming it yourself, because nothing inside the file records which account or which date it came from beyond what you note down.
A live stream cannot be downloaded the way a finished post can, because the file does not exist yet. It has to be captured as it airs, so recording runs as a job rather than as an instant download: the job is queued, the recorder attaches to the stream, and it keeps writing until the broadcast ends or you stop it. Jobs report their state as queued, recording, completed or failed, and finished recordings expire after a retention window instead of being kept indefinitely. Because the work outlives the request and runs on the server, live recording uses an account, unlike the single-video form.
A broadcast is served as a segmented stream rather than as one file, which is why recording means collecting segments for the duration and assembling them afterwards. Long streams therefore produce large files, and the recorder keeps writing in the background whether or not your browser stays open. Jobs are queued, so a capture can start while another is still running, within the limits of the tier the account is on.
What the recorder does
Captured while it airs
The recorder attaches to the running broadcast and keeps writing for as long as it lasts.
Stop it when you want
A running job can be stopped early, and what was captured up to then is kept.
Runs server side
You do not have to keep a tab open or a phone awake for the length of the broadcast.
A state you can check
Queued, recording, completed and failed are reported per job, so a long capture is not a black box.
Record a live stream
- 1
Find a broadcast that is live
Open the profile while it is streaming and copy the link to the live stream.
- 2
Start the job
Submit the link on the live recorder page. The job is queued and the recorder attaches to the stream.
- 3
Collect the recording
When the broadcast ends, or after you stop the job, the finished file is available from the job entry.
Limits worth knowing
A broadcast that is already over leaves nothing to record. Capture begins at the moment the recorder attaches, so whatever happened before that is not in the file. A stream that is private, region blocked or taken down mid-broadcast will cut the job short, and a job that fails reports that state rather than returning a file. Recordings expire after a retention window, so a capture you want to keep should be downloaded.
FAQ
Can I find and download a broadcast from last week?
No. Only a stream that is airing can be captured, and an ended broadcast is not retrievable.
Is my archive stored on Savett?
No. A completed recording expires about an hour after the job finishes, so it has to be moved to your own storage.
Can I download a live stream that already ended?
No. Nothing is recorded in advance, so a broadcast has to be captured while it is running. Once it is over there is no file to retrieve.
Do I need an account for live recording?
Yes. The recording runs on the server for the length of the broadcast and belongs to a job you come back to later, so it is tied to an account, unlike a single video download.
What happens when the streamer ends the broadcast?
The job completes on its own and the recording up to that point is kept, so you do not have to be watching when it stops.
Is the whole broadcast in the file?
Only from the point the recorder attached. A stream that was already an hour in when the job started gives you the remainder, not the beginning.
What file do I get from a recording?
A single video file assembled from the segments captured while the recorder was attached, downloadable from the job entry once the job reports as completed.