GUIDE深度 · Deep dive

Speech Recognition Privacy

The Web Speech API is only a browser interface; recognition may happen in the vendor cloud. Do not use it for meetings, client calls or medical data without checking compliance; use a local model or a contracted service instead.

更新 2026-11-01 · 公式可核验

Chrome speech recognition often relies on the cloud.

Once microphone permission is granted, the page can keep capturing.

Use local Whisper-like models for sensitive content.

Close the tab and revoke microphone permission afterwards.

FAQ

Does the browser save audio?

Depends on the vendor; the page cannot guarantee it, so treat it as untrusted.

Does it work offline?

Usually not; cloud recognition needs a network.

Sources

Related calculators

Speech to TextTranscribe Chinese/English live with browser speech recognition, copy or download TXT; requires Chrome/Edge and microphone permission.

More guides

Dictation WorkflowRead guide
KEEP READING

继续读

按顺序读,或跳到同主题的其他指南。

全部指南 →