Voice to text

How it works

1. The page uses the browser’s built-in SpeechRecognition API. Recognition runs continuously and restarts itself if the browser cuts the session short.
2. Interim results appear in grey under the box; only finalised phrases are written into the editable text area, so you can keep typing corrections while you speak.
3. The tidy-up option trims stray leading spaces and spaces phrases apart. It does not rewrite your words.
4. The counter switches to characters for Japanese, Korean and Chinese, where counting whitespace-separated words is meaningless.

What it will never do

• It does not promise on-device recognition. Most desktop browsers, Chrome in particular, send the audio to the browser vendor’s servers to be transcribed — that is the browser’s behaviour, not a choice this page can make.
• It never stores or transmits anything itself. The text stays in the page and is gone when you close the tab.
• It does not identify speakers, add timestamps, or transcribe an audio file — it listens to a live microphone only.