wc-seed-podcast

To transcribe podcasts for seed data, this Skill queries Listen Notes for relevant episodes, matches audio to YouTube, and uses seed workers to generate transcripts stored under seed/podcast. This builds a diverse text library for AI-assisted writing workflows.

You discover episodes by providing a queries JSON file containing podcast_searches. The Skill queries Listen Notes across multiple pain_point and sub_question angles, extracting episode metadata and filtering results to ensure strict alignment with your seed-phase content curation goals.

The best way to filter and deduplicate podcast data is to apply strict duration constraints between 10-90 minutes and remove duplicate entries by URL. This ensures only diverse, high-quality audio sources are preserved for transcription and downstream seed data compilation.

Yes, you can use YouTube matching to transcribe podcast audio automatically. The Skill discovers episode metadata via Listen Notes, finds corresponding YouTube audio, and delegates the actual transcription process to automated seed workers without requiring manual file downloads.

You need to provide a queries JSON file structured with a podcast_searches array. This file serves as the input for the seed-worker, directing the Listen Notes queries to discover, filter, and transcribe podcast episodes into the designated seed dataset.