Description
Description
Build a newsroom-ready pipeline that ingests daily broadcasts and turns them into searchable stories. You will learn how to schedule batch jobs, manage hot folders, and normalize mixed sample rates from field recordings. We demonstrate diarization that distinguishes host, guest, and reporter while handling phone-in audio. A module on metadata shows how to attach program, segment, and rights information to each transcript slice. You will implement quality gates that quarantine low-confidence audio for review without stopping the whole run. Examples show how to auto-generate social clips by cutting around keyword hits with speaker context. We cover content safety: profanity flags, name redaction options, and jurisdiction-aware policies. You will connect the output to a hybrid index so keyword and semantic queries can both find relevant moments. Dashboards visualize throughput, failure causes, and per-show accuracy trends so editors see health at a glance. By completion, your team can archive, search, and share clips within minutes of air time using a repeatable process.
Format
Step-by-step videos, DAG templates, sample dashboards, policy checklists, starter hybrid index config
Duration
5 hours including implementation labs
What You’ll Learn
– Broadcast ingest patterns
– Diarization for roles
– Rights & show metadata
– Confidence gating
– Clip auto-generation
– Hybrid search wiring
Target Audience
Newsrooms, public media archives, and comms teams processing recurring shows
Videos + DAG templates + dashboards + policy checklists + index configs
5 hours
– Ingest patterns
– Role diarization
– Metadata & rights
– Confidence gates
– Auto clips
– Hybrid search
Newsrooms, media archives, communications teams
English