Why Did ASR Create A Draft When Audio Quality Is Poor

This article explains how ASR determines when to produce a draft.


The system has been producing drafts for some time then the clinician’s audio quality degrades. If the audio quality becomes poor, why does the system continue to produce drafts?


When new drafts are edited and sent back to ASR for analysis, the analysis data is kept in historical database. This data is retained for a period of time but is continuously refreshed as new drafts are analyzed. As new data is added to the database some of the older data is released.
ASR uses the historical data to learn from and predict output on future drafts. When the audio quality degrades on new jobs, the system continues to produce drafts because historically the draft quality was more efficient to edit than transcribe. If the system continues to get jobs with poor audio quality the history would eventually change and ASR would stop producing drafts and place the clinician back into transcription status. See workflow below: