New
- Transcription queue depth endpoint
- New unauthenticated endpoint
GET /api/status/transcription-queuereturns{"pending": N}— the number of jobs currently waiting in the in-memory transcription queue - Queue depth also included in the existing
GET /api/status/performanceresponse astranscription_queue_depth
- New unauthenticated endpoint
Fixes
- Tone-detection talkgroups correctly bypass minimum call duration for all calls — short calls on tone-detection-enabled talkgroups were incorrectly being skipped; they now always proceed to transcription as intended
- Duration check uses original audio —
getCallDurationnow probesOriginalAudiofirst and falls back to converted audio, so servers without audio conversion get accurate duration measurements