Technology
ASR (Automatic Speech Recognition)
The speech-to-text foundation under every ambient scribe.
ASR is the technology that converts spoken audio to text. Every ambient scribe is built on an ASR layer plus a language-model layer that turns the transcript into a structured note.
ASR quality is *not* the same as note quality. A scribe with perfect ASR can still hallucinate in the note-generation step; a scribe with imperfect ASR can still produce clean notes if the language model handles ambiguity conservatively. Don't conflate the two.
See also
- Speaker diarization — Separating who said what in a multi-speaker recording.
- AI hallucination — When an AI model writes something confidently into a note that was never said.