Why Does My Product Video Start With the Wrong Subtitle Language?

A correct subtitle file does not choose its own display state

An HTML video can contain multiple track elements. Each can declare a source file, a kind, a language through srclang, a human-readable label and an optional default attribute.

MDN documents that the default attribute identifies a track to enable when user preferences do not indicate a more appropriate choice. Browser and player behavior can also depend on user preferences and application code. Therefore, a language opening by default is not always evidence that the caption files were exported incorrectly.

Keep the language of the spoken audio separate from the language of the captions. A translated promotional overlay burned into the picture is not a selectable subtitle track. Likewise, a file named zh.vtt does not automatically become a Chinese track if its HTML metadata declares English.

Inspect metadata and actual selection

Ask the maintainer to inspect the rendered video element, not merely the original template. For each track, record the src, kind, srclang, label and presence of default. Confirm that the WebVTT file is reachable and contains the approved language and timing.

Then examine the active text track in the actual player. If the English file is attached to a Chinese language label, repair the metadata association. If the correct Chinese track exists but English is active, inspect default selection, saved preferences and application-level player logic. If the player exposes no language choice, investigate its implementation rather than assuming every native or embedded player provides the same controls.

Do not add default indiscriminately to every language. A single intended default is easier to reason about, but the actual selection must still be checked under the target browser's preference settings.

Hypothetical case: the files are correct but the default is wrong

Hypothetical illustration, not a browser test: A fictional product demonstration has approved English and Simplified Chinese WebVTT files. The English file is associated with srclang="en" and the Chinese file with srclang="zh-CN". Assume the intended review environment has no overriding caption-language preference.

The page marks the English track as default, although the merchant's approved localized page should initially present Chinese subtitles. The proposed correction is to change the intended default selection in the webpage configuration, retaining both language-specific files and their accurate metadata.

If the player still opens English after the change, the next investigation is actual player state and preferences, not another translation. The worksheet distinguishes the intended language from the active track; it does not claim a completed playback result.

Verify the destination, not just the HTML

After a supported implementation change, load the final video in the intended browser and device. Check the initial language, switch to the other language and return. Confirm that the selected track's words match the final audio, product model names and edit timing.

Repeat with a clean preference state where practical, and document any player-specific language behavior. A page that displays the correct default but prevents users from selecting the other approved language has not satisfied a two-language delivery requirement.

The existing two-language subtitle export guide addresses missing sidecar files. This article starts after the language files exist and examines their HTML selection metadata. Subtitle installation and playback configuration happen in the external website or video platform, not in CartVisual.

Record your own observations

The filled rows are illustrative examples. Use the blank row for your own files and observations.

Download the editable CSV worksheet

What the references establish

• MDN — track HTML element: Documents kind, srclang, label and default attributes, and the relationship between text tracks and language selection.
• MDN — TextTrack mode property: Defines disabled, hidden and showing states for a text track, supporting inspection of the active track independently of its file contents.