Ermine.ai

Ermine.ai records from your microphone and transcribes speech entirely inside the browser, so no audio is sent to a server. It suits people who want quick English transcripts without uploading sensitive recordings.

On first use the page downloads a speech recognition model of about 50 MB and caches it, which can take a few minutes. Later sessions start much faster. After you grant microphone access, you click to begin and the transcript builds up as you talk.

When you are done you can download both the audio recording and the text. Transcription currently supports English only, and there is no account, upload or file import step.