Record where the sound actually goes
Open Windows sound settings and identify the default output, then note the device carrying the programme. A television connected by HDMI, a USB headset and laptop speakers are distinct endpoints. Keep the application name with this record, especially if it has its own output selection.
An illustrative mismatch is an application deliberately routed to a headset while the system default remains a television. Do not assume this example describes your computer. Compare the actual selected endpoints before changing them, and save the original values so you can restore the preferred listening setup.
Use one spoken sample
Confirm that Live captions has completed language setup, then play a familiar spoken passage at a normal level. Follow Microsoft’s guidance for aligning the relevant default output and repeat the same passage. Avoid mixing several simultaneous applications with speech, since that makes the result harder to interpret.
Record whether text begins, remains blank or represents the wrong language. An empty caption box is a different outcome from incorrect automatic wording. Changing the translation target will not establish that the feature is receiving the intended sound in the first place.
Keep capture and privacy choices deliberate
Do not enable microphone capture merely as a workaround for a missing playback path without considering what room audio it may include. If a microphone is needed for a separate in-person conversation task, review its permissions and clearly separate that task from programme audio.
When seeking help, report the default output, application output where available, caption language and the result of the single-sample comparison. Use a neutral sample instead of sharing private conversation audio. A working headset proves the listening route, while a functioning caption stream is a separate result that should be recorded accurately.