Quick answer: Record 30 min, upload, AI processes 5 min (transcript + noise removal + levelling + chapters), review 10 min, add title/artwork 5 min, publish to all platforms 5 min. Total: ~55 minutes for a mid-complexity episode.

Minute 0–5: Setup

Check microphone levels (peak around -12dB), set sample rate (48 kHz), disable notifications, close the door, put your phone on silent. On Mac: enable “Use this device for sound output” on the correct audio interface. On Windows: disable audio enhancements for the input device.

Minute 5–35: Record

Record straight into a cloud recorder (GeraCast, Riverside, or even QuickTime). For interviews, record each person locally (48 kHz WAV) and sync after. For solo, a single stereo WAV is fine. Do a 5-second silent room-tone capture at the start and end — AI noise-removal uses this to profile the room.

Minute 35–40: AI processing

Upload to GeraCast. AI runs five jobs in parallel:

Transcription — Whisper-class model, speaker-labelled, 40+ languages
Noise removal — background hum, HVAC, keyboard clicks
Levelling — loudness-normalised to broadcast -16 LUFS
Chapter detection — topic boundaries with titles
Show-note generation — summary, bullet points, guest bios

This typically finishes in 5 minutes for a 30-minute episode.

Minute 40–50: Review

Scrub the AI transcript. Fix names, technical terms, and any hallucinations. The AI chapter detection is usually 90% right — you'll tweak one or two boundaries. Check levelling (it should sound even across the whole episode). Delete the room-tone captures.

Minute 50–55: Metadata & artwork

Episode title — keyword-first (“How AI Hiring Works in 2026 — with Jane Smith” beats “Episode 42: Jane”).
Episode summary — AI-drafted, human-edited. 2–3 sentences above the fold.
Chapters — AI-generated; review.
Artwork — 3000x3000 px square, <500 kB, RGB (not CMYK), JPG or PNG.
Transcript — published as a separate page for SEO and accessibility.

Minute 55–60: Publish

Click publish. The RSS feed updates; Apple Podcasts, Spotify, and YouTube Music pull within 10–60 minutes. GeraCast also pushes a short-form video clip (3 highlights) to your social queue for manual review.

What takes longer

Narrative/scripted shows still need human editing: removing fluff, tightening transitions, SFX. Expect 2–4 hours of post per 30 min of finished tape. AI gets you most of the way for interview and conversation formats, not for narrative.

The One-Hour Podcast Workflow: From Recording to Published in 2026