How to automate podcast transcription with AI pipelines
How to automate podcast transcription with AI pipelines
Producing a podcast episode generates 30-60 minutes of audio, but the derivative content — transcripts, show notes, social clips, blog posts — takes hours of post-production work. An automated podcast transcription pipeline converts audio to text, generates summaries, extracts key quotes, and publishes everything to your CMS and social channels. Post-production time drops from 3-4 hours to under 15 minutes.
What the pipeline should produce
Full transcript — timestamped, speaker-labeled text. Show notes — 200-300 word summary with key topics, guest bio, and links mentioned. Key quotes — 5-8 quotable snippets for social media and newsletter highlights. Chapter markers — topic-based timestamps for Apple Podcasts, Overcast, etc. Blog post draft — 600-800 word article derived from episode content, optimized for SEO.
Building in CodeWords
Tell Cody: "When an audio file is uploaded to a specific Google Drive folder, transcribe it using Whisper, identify speakers, generate show notes and key quotes using Claude, create chapter markers, write a blog post draft, and publish everything to our Airtable CMS. Post the show notes summary to #podcast in Slack."
Cody scaffolds: (1) File watcher monitoring a Google Drive folder for new uploads, (2) Transcriber sending audio to Whisper returning timestamped text with speaker diarization, (3) Content generator sending transcript to Claude with five prompts (show notes, key quotes, chapter markers, blog post draft, social posts), (4) Publisher writing all outputs to Airtable, uploading transcript to Google Drive, and posting summary to Slack.
Ensuring accuracy
Raw Whisper output averages 90-95% accuracy. For published content: run an LLM correction pass (send transcript to Claude to fix transcription errors without changing the speaker's meaning), provide custom vocabulary (domain-specific terms, product names, guest names), and flag low-confidence segments in Google Sheets for human spot-check. Zapier and Make can move files but can't transcribe audio or generate show notes. n8n has file handling but no native speech-to-text or LLM processing. CodeWords runs the full audio-to-published-content pipeline.
