mcardle-features-dashboard - #13
Conversation
* Wire the transcribeme db schema information into the dashboard * Introduce reusable aggregate view of tables * Wire reusable aggregate view into each page
* Introduce the pipeline-failures page * Wire the transcribeme db schema information into the dashboard
Allows a user to explore and override audio qc failures.
* Introduce datetime parse/missing interview reconciliation page * Runsheet match page into nav sidebar * pm2 configuration and directions * Surface runsheet redcap event name in matching view (unmatched runsheets only) * Override audit trail page for visualization
|
Posting @rtmcardle 's comments here for the record:
|
| if (!status || !status.wav_conversion) { | ||
| return ( | ||
| <div className="m-16"> | ||
| <Empty description="This recording has not entered the AMPSCZ transcription pipeline yet" /> |
There was a problem hiding this comment.
nitpick: We should use consistent branding for the pipeline as dpinterview
dheshanm
left a comment
dheshanm
left a comment
There was a problem hiding this comment.
This PR adds:
- New issue types: AudioQC failures, awaiting vendor transcription, pipeline failures
- Aggregation summaries for most issues, grouped by various dimensions
This significantly improves visibility into the transcription pipeline, surfacing:
- QC failures (files not sent for transcription) — with support for manual overrides to force transcription
- Files waiting on the vendor
- Files not yet pushed to the vendor
Implementation looks solid, with parameterized SQL queries on TranscribeMe model. Approved!
non-blocking nitpick: I would lean toward computing hours_waiting client-side rather than in the query — keeps the SQL deterministic and cacheable. Fine as-is though.
2 participants
Additional features developed in July 2026, surfacing TranscribeMe data, pipeline failures, and linking runsheets to malnamed files.