ASR Rejected – Scenarios and Troubleshooting


This article explains what the “ASR Rejected” status means in the activity log.

When looking at the activity log in InCommand for a particular TID, you might come across the “ASR Rejected” entry.
This can be in either outstanding jobs or delivered jobs. The screenshot below shows after the job was delivered.

Activity22.png


“ASR Rejected” means a draft was produced but the quality was too low to be edited. So the draft is “rejected” and sent
to typing status.


ASR Rejected Scenarios

1. The clinician is new and the system determines enough basic data exists that a generic speech model can be used for
drafts. The generic speech model is not as individualized as the user specific model so quality tends to be lower and
more jobs get rejected. While the generic model is used, the user specific model builds in the background. When the
user specific model is complete it is used from that point forward to create drafts.

2. The system has been producing drafts for a period of time but something changed in either the clinician’s audio,
environment, or speaking style, which caused the draft quality to drop. The reduction in quality results in draft rejection.
Note for this situation to occur, there has to be a consistent change in a large number of jobs processed that affect the
historical model database.

For more information, see the article titled Why_Did_ASR_Create_A_Draft_When_Audio_Quality_Is_Poor

Troubleshooting “Rejected” Status

Seeing rejected jobs is not necessarily a bad thing. Often times jobs are rejected when the clinician is new and
the generic speech model is used. As the system processes more speaking and transcription data, the number of
rejected jobs should decrease.

If you’re noticing a clinician is consistently getting rejected after a reasonable amount of time (after approximately 100
documents and 120 mins of audio) please contact your support team for further investigation.