Find which control owns the text
Text over a video can come from the programme's caption track, the website's player or a browser or operating-system speech-recognition feature. These layers may have separate controls and language choices. If turning captions off in the player leaves a floating text box visible, investigate the browser or system feature before concluding that the player's button is broken.
Chrome's Live Caption is a documented example of automatically generated speech captions. Google's guidance distinguishes local Live Caption processing from translation features that can send captions to Google. Check the specific feature and its current privacy explanation rather than assuming every text or translation feature has identical processing. Availability also depends on the actual device and supported languages.
Use a short two-control comparison
In a hypothetical lesson, the player offers an English caption track while the browser also generates English text. Both appear at once. The viewer first disables only the browser feature and confirms that the publisher track remains, then tests the player control. This establishes which interface owns each text layer without changing the audio or signing out.
Record the location and appearance of the text as well as the control used. A movable browser caption bubble may be visually different from captions rendered inside the video. Do not rely on appearance alone, however; the decisive observation is which documented setting changes it. Take care that a screenshot does not expose a private lesson title or account information.
Report errors to the right place
An automatic transcription error and a mistake in a publisher-supplied track need different reports. For the publisher track, include programme, language and approximate time. For an automatic feature, include the browser or system version, selected recognition language and a brief authorised example. Avoid sending an entire confidential conversation just to demonstrate one misheard word.
Captions may need speaker identification and meaningful non-speech information as well as dialogue. Automatic speech text should therefore be assessed for the actual viewing need rather than assumed equivalent to a carefully prepared accessible track. If a sound effect carries essential information, check whether the chosen text communicates it. Ask the publisher for the available accessibility provision when this matters.
Confirm the everyday configuration
Once the layers are understood, select the arrangement you can reliably read and operate. Test it in the mode you use, including fullscreen or a floating window if relevant. A feature working on one page does not establish how every website or player will expose its captions. Keep the controls you need reachable.
Write the result in practical terms: the player track provides the required language, automatic captions are disabled to avoid duplicate text, and the chosen display remains readable. If the required track is absent, say that directly. A browser feature may help a particular session, but its presence does not prove that the publisher provides the requested language or accessibility track.
Questions answered
Why does text remain after I turn off the player's captions?
A browser or operating-system caption feature may be generating a separate layer. Check which documented control changes the visible text.
Are automatic captions always equivalent to a publisher's accessibility track?
No. Assess accuracy, speaker identification and meaningful non-speech information for the actual content and viewing need.
Sources & further reading
Official references for the subjects discussed. Service terms and device software can change.