Voice systems · browser experiment
SpeechCapture
A small browser laboratory for testing multilingual speech recognition, microphone events and the transcript returned by the Web Speech API.
01 · INPUT
Choose what the browser should listen for.
The interface language and recognition language are independent. Choose the locale spoken into the microphone, then start a single recognition session.
MicrophoneCHECKING
Microphone handling depends on the browser and its speech-recognition service. This page does not implement its own audio upload or recording store.
02 · TRANSCRIPT
What the engine heard.
The final recognition result appears here. Confidence is reported when the browser exposes it.
Latest result--:--:--
No speech captured yet.
Confidence—
Words0
Localept-PT
03 · EVENTS
Recognition diagnostics.
The event stream below is intentionally raw. It is useful for distinguishing microphone activation, sound detection, speech detection, recognition results and browser errors.
