We love listening to podcasts, but it is hard to find specific quotes, interview guests or chapters within an episode. We built Tapedeck to make a democratic and accessible form of online journalism and entertainment even better for listeners, creators and advertising partners.

What it does

Tapedeck transcribes podcasts in the cloud—text is much easier to work with than audio. Using machine learning and heuristics, it creates chapter markings, extracts key quotes and identifies guests.

Our technology also helps listeners find new great content. It identifies themes in chapters with keyword frequency analysis and suggests thematically similar shows.

Tapedeck drives listener engagement and connects podcast creators and journalists to new audiences.

How we built it

We built Tapedeck with React Redux—the frontend runs entirely in your browser, across all devices. Our backend is written in Python and hosted on an AWS EC2 instance.

Challenges we ran into

We wanted to improve the presentation of our exploded view episode display, but due to time constraints we had to focus on our backend. Detecting pauses in speech that may indicate a new thematic chapter was especially tricky, but we managed to do it.

Accomplishments that we're proud of

We think Tapedeck works and looks amazingly well for something that's been knocked up in just over two days.

What's next for Tapedeck

We want to improve both user experience and backend—there's quite a road ahead of us.

Share this project: