Browser speech recognition is not reaching its service.
This is a network problem, not a microphone problem — your mic is fine.
Common causes: an ad blocker or privacy extension filtering
*.google.com/speech-api, a network or DNS that blocks Google,
or a Chromium/Brave build that ships without the API keys Chrome's
recogniser needs.
Fix: switch Recognition engine to
On-device Whisper. That model runs entirely inside this page —
free, no key, no Google, and it works offline once downloaded.
Transcript
0 words0 characters
Limit 12,000
What changed
0 fixes
Translation
MYMEMORY
This browser cannot dictate.
Speech recognition needs Chrome, Edge, Safari or Opera. Firefox does not
implement it. Everything else — correction, translation, copy, download —
still works, so you can type instead.
Microphone blocked.
Browsers only allow microphone access over https:// or on
localhost. Serve this app over HTTPS to enable dictation.
How your text is handled.
With Local whisper.cpp and On-device Whisper, your audio
never leaves this machine — transcription runs here or in this page.
With Browser speech, dictation is sent to your browser vendor's
recogniser, which in Chrome and Edge means Google; that is why that
mode can fail on locked-down networks. Spelling and grammar fixes always
run on this server and never leave it. Only translation sends text, and
only to MYMEMORY,
which stores a hashed copy of each request for caching.