Tapesearch

Tapesearch is an AI search tool that searches podcast transcripts and jump straight to the moment that matters. A podcast transcript search engine covering millions of episodes, with timestamp jump links, AI chat over conversations, keyword alerts, downloadable transcripts and an API plus MCP server for assistants.

Tapesearch is a search engine built on a large open database of podcast transcripts, currently holding more than five million transcribed episodes. Instead of listening through hours of audio to find a single remark, you type a phrase and get back the exact episode and timestamp where it was spoken. The index leans on the most popular shows by Apple Podcasts rating volume, with the stated long term aim of widening coverage. Transcripts are refreshed on a daily cycle starting at midnight UTC, and shows that publish through the Podping notification system are typically transcribed within minutes. The product serves two fairly different audiences. Curious listeners, journalists and academics use the search interface to pull timestamped quotes, read instead of listen, and download transcripts as TXT, SRT, VTT or JSON. Market researchers, brand teams and open source intelligence analysts lean on alerting and trend charting to watch how often a keyword, brand or advertising pattern surfaces across thousands of conversations. Podcasters use the same index to mine their own back catalogue and to publish transcripts for discoverability and accessibility. Beyond the web interface, Tapesearch exposes search and transcript download endpoints through an API, and hosts an MCP server so assistants such as Claude can query the database without any local installation. Chat style querying lets you ask questions across multiple shows and receive consolidated answers rather than a list of links. Accuracy is worth setting expectations around. The transcription model handles ordinary speech well but slips on names, rapid delivery and speech layered over music, and long silent or musical stretches can push timestamps slightly out of sync. Dynamic ad insertion is another source of drift, since the spot you hear may not match the one captured at transcription time. Paywalled and exclusive shows, Spotify among them, are not yet indexed. Against general podcast players and manual transcript reading, Tapesearch competes on speed of retrieval and on the API and alerting layer rather than on playback.