StreamTranslateCompare against the YouTube video
Watch the stream’s audio and captions together in the YouTube player. Comparing captions with your own voice in the room also includes YouTube’s video delivery delay, which is a different measurement.
Use three short phrases with pauses, then a longer passage. Note whether captions lead or follow the matching words in the player and whether the gap grows during continuous speech.
Separate the stages
| Stage | What it affects |
|---|---|
| Recognition | When spoken words become text |
| Translation | When the chosen translation is ready |
| Caption posting | When completed text reaches YouTube |
| Video delivery | When viewers hear the matching audio |
A completed sentence can arrive later than an interim word preview. A timing preset does not speed up recognition or translation.
Use Advanced only for a consistent early feed
If captions consistently precede the matching player audio, Control’s Advanced caption timing can shift them 1–5 seconds later. Choose the setting before enabling captions. It cannot make late text arrive earlier or repair an increasing backlog.
Record a useful timing report
Report the chosen languages, Studio latency setting, timing preset and a short private recording showing both captions and player audio. Include whether the issue appears at startup, after reconnect, or only during long speech. Mask all keys and caption URLs.
Example: leading text versus a growing queue
If every line appears about two seconds before its matching player audio, a later timing preset may improve alignment. If the first line matches but each following line falls farther behind, adding two seconds is the wrong change: it introduces more delay. Note the exact phrases and whether text arrives in a burst after a pause. That distinction helps separate a constant alignment problem from recognition boundaries, translation delays or a posting queue. Make one change and repeat the same passage.
Sources and setup limits
See YouTube’s official requirements for the underlying caption transport and latency modes. Actual behavior must also be checked on your own broadcast.
Common questions
Does a 20-second stream delay mean captions are 20 seconds late?
No. Measure captions against the audio inside the same YouTube player. Delivery delay from your room to that player is separate.
Should I increase timing when captions are late?
No. The later preset is for captions that lead the matching audio. Adding delay does not fix late text.