# Podcast Transcript Search: Build a Searchable Episode Archive Quickchat has a [tutorial for building a searchable podcast archive]() in [Quickchat AI](). You ask for an idea you half-remember, such as a guest's advice about leaving silence after a question. The assistant answers with the guest, the episode, a supporting quote, and a timestamp when the source has one. The walkthrough starts with a transcript export from Riverside's edited recording. Open the transcript panel's three-dot menu, choose Download transcript, pick Transcript (.txt), and leave "Include cut content" off so the archive can't return a quote that never made it into the published episode. Riverside's TXT has speaker labels and a timestamp at each speaker change, so you get a source position rather than one for every sentence. Transcript downloads need an eligible paid plan and an account role with download access. The tutorial also covers where to find text in Descript, YouTube Studio, Spotify for Creators, and Apple Podcasts Connect. It doesn't transcribe audio, so you need a transcript before you start. Before importing, you add a short header to each file with the show, episode number, title, guest, canonical episode URL, and whether the text is a full transcript or an excerpt. That header is how the assistant attributes advice to the right person and episode. Quickchat's .txt upload limit is 1 MB per file. For longer transcripts, split at topic boundaries and repeat the header in each part. Then add the files to the agent's Knowledge Base and wait for each to show Ready. The example is a fictional interview show, Second Draft, with three short excerpts. A starter kit includes the sample sources, episode inventory, Main Prompt, and a test worksheet, so you can reproduce the exercise before using your own episodes. Loading replies --- Canonical source: https://creatorstoolbox.com/@scout/post/n_2sj39f1mkoku9