Podcasts

FeynmanLM can extract transcripts from Apple Podcasts episodes, making your podcast listening available as sources for AI-assisted learning.

Apple Podcasts only

Transcript extraction currently only works with Apple Podcasts. Apple Podcasts stores transcripts locally on your Mac, which FeynmanLM reads directly. No third-party API involved.

How Episodes Are Discovered

FeynmanLM queries your local Podcasts database for episodes you've recently played (last 90 days) and checks whether a transcript is available.

This requires Full Disk Access in macOS System Settings → Privacy & Security → Full Disk Access (toggle FeynmanLM on).

Transcript Availability

Apple Podcasts downloads transcripts on demand. They are not available by default. To make a transcript available:

1. Open Apple Podcasts on your Mac.
2. Navigate to the episode.
3. Click Transcript in the episode detail view. This triggers Apple to download and cache the transcript locally.

Once cached, FeynmanLM can extract it. Episodes without a cached transcript are skipped.

Transcript Cleanup

Raw auto-generated transcripts are messy: no paragraph breaks, no speaker labels, filler words everywhere. When you have an OpenAI API key configured (Settings → API Keys), FeynmanLM automatically cleans up transcripts using a lightweight LLM call (GPT-4o-mini):

• Adds paragraph breaks at natural topic transitions
• Adds speaker labels (Host:, Guest:) when speaker changes are detectable
• Removes filler words (um, uh, like, you know) that add no meaning
• Fixes punctuation and capitalization errors from auto-transcription

The full transcript is preserved — nothing substantive is removed. This same cleanup runs for YouTube video transcripts.

Using Podcasts with AI

Once transcripts are extracted, they appear as sources in Studio. Select them and your AI assistant can:

• Summarize key points from the episode
• Quiz you on specific claims or mechanisms discussed
• Compare ideas across multiple episodes or with your other sources

Tips

• Episodes with long-form interviews and technical discussions work best.
• Make sure to open the transcript view in Apple Podcasts before trying to extract.
• Podcast sources use local file paths, ensuring they're tied to the actual transcript you listened to.