podcast-dig

Exact quotes from long podcasts, searchable in my Obsidian vault.

Pull an episode

/podcast-dig with a company name finds its founder and investor interviews, ranks them, and files each episode in my vault with a full, timestamped transcript.

podcast-dig Library in Obsidian: starred episodes grouped by company, such as Granola with Chris Pedregal on AI & I, each tagged Transcript
Filed by company: Granola

The transcript follows playback

Press play and the reader scrolls with the speaker, so the line being said is always on screen.

podcast-dig reader: Ben Thompson's paragraph at 27:43 highlighted as it plays, above a player bar with Follow turned on
Follow keeps the playing line in view

Keep a line, with its timestamp

Highlight a line and it goes to My takeaways with its timestamp, the speaker's exact words and my own note.

My takeaways card: timestamp 03:40, my note, and the exact quote about US dependence on China
The exact words, kept at 03:40

Notes in plain Markdown

Kept lines land in an ordinary Obsidian note: timestamp, exact quote, my note and a link back to the episode.

Obsidian note titled 20VC The $100 Billion AI Assistant Race, Town: timestamped quotes from JD Grèze, each with a short note and a link to the episode
Timestamp, quote, my note, source link

One cache shared with LexiAI

podcast-dig checks corpus.db, the local Whisper cache it shares with LexiAI, before transcribing, so each episode is transcribed once on my Mac. LexiAI's podcast clips come from the same cache.

LexiAI Heard in podcasts for appall: a Diary of a CEO line at 25:35 with the word highlighted and a play button
Same cache, as a word clip in LexiAI

It also supplies the raw material for Startup Research: founder interviews, quoted exactly.

Built with Python, local Whisper, RSS and Obsidian, as a Claude Code skill · Designed and built by me

Contact me

qing.mao.2024@marshall.usc.edu