How to Turn a Podcast Episode into Show Notes with Castmagic

November 19, 2026

Castmagic takes an audio or video file, transcribes it, and generates show notes, timestamps, pull quotes, and social captions from templates you define. A 60-minute episode takes about five minutes to process and another fifteen to fix.

That fifteen-minute editing window is realistic. Expecting zero edits is not.

What do you need before uploading?

Three things, in this order.

  1. Clean audio. Export your episode mix, not the raw multitrack. Castmagic handles room noise, but two overlapping mics confuse the speaker diarization.
  2. A glossary. Its custom vocabulary field takes names, product terms, and acronyms. Without it, “Kubernetes” becomes “cooper netties” and “Anthropic” becomes “anthropic” — spelled wrong in every downstream asset.
  3. A template. The built-in outputs are generic. A custom template with your section headings produces something that actually matches your site.

Skipping the glossary is the single most common cause of a bad first run. Fill it before the first upload, not after the third.

How do you set up a template that works?

Open templates, create a new output, and specify fields with plain instructions.

For a typical interview show:

  • Summary: 80 words, third person, no marketing language
  • Key points: five bullets, each with the guest’s name
  • Chapters: timestamps every topic change, titled as questions
  • Quotes: three pull quotes, 15–30 words, speaker attributed
  • Social: one LinkedIn post under 200 words, one thread of four posts

Write the instructions as constraints, not requests. “80 words” beats “a concise summary” because the model can check a count and cannot check a vibe.

Why are the chapter timestamps wrong?

Because chaptering is the weakest part of any transcript-based tool, and Castmagic is not exempt.

Two failure modes. First, timestamps drift on long episodes — a chapter marked at 14:20 may actually start at 14:48. Second, the chapter titles describe what the model thinks the topic was rather than where the conversation actually turns.

The fix: always play the episode while checking chapters. Fifteen minutes of listening, scrubbing to each boundary. Then correct the timestamps by hand.

For a 60-minute show, expect two to four wrong boundaries. It is faster to fix them than to explain to a listener why the link lands mid-sentence.

The transcript viewer helps here. Word-level timestamps are clickable, so you can find the real start of a segment and read the exact time off the pane.

Which export formats matter?

Output Where it goes Watch out for
Show notes Podcast host description field Some hosts strip line breaks
Chapters Apple Podcasts, YouTube Must be ascending and start at 00:00
Transcript file Accessibility, SEO page Raw output has filler words
Social captions LinkedIn, X Character limits differ per network
Blog draft Website Needs a real edit, not a paste

Apple Podcasts requires chapters formatted in its own spec — a text list with timestamps. Others accept YouTube-style chapter lines inside the description. Check which your host wants before exporting.

How much does Castmagic cost?

It is credit-based. Transcription minutes and generated outputs both consume credits, and plans run anywhere from roughly $20 to $100+ a month depending on volume. A weekly hour-long show sits comfortably in the middle tier.

What eats credits unexpectedly: regenerating an output. Every retry costs. That is another reason to fill the glossary first and get it right in one pass.

Where it does not fit

Shows with heavy crosstalk between four or more speakers will get speaker labels wrong, and correcting them manually across an hour is brutal. Highly technical episodes with dense jargon need so much glossary work that the time saving shrinks.

Also: if your show is scripted, a transcript-based summary tool adds less than simply editing your own script into notes. The value is in unstructured conversation.

Quick Summary

  • Castmagic transcribes audio or video and generates show notes, chapters, quotes, and social posts from templates.
  • Fill the custom glossary before your first upload or names get mangled across every output.
  • Expect two to four wrong chapter timestamps per hour — verify by scrubbing the audio.
  • Credit-based pricing roughly $20–$100+ monthly; regenerating outputs costs credits.
  • Works best on two-person unscripted conversations; struggles with four-plus speakers.