Spinning Yarns 01

Introducing Yarn, super short video clips for chat. I’m Jeffrey, a founder, along with Chris Butler, we came up with Yarn because be needed better ways to search inside of videos across different dimensions.

We’ve ingested and analyzed many (many, many, many) movies, tv shows and music videos. In this series of blog posts I’ll reflect on some of the technical and design challenges, and give some insight into more of our data, statistics and processes.

First off, take a look at our site: Yarn, you’ll see a series of popular clips. Click on any clip…it should be super short probably less than 3 seconds (our average length clip is 2.887 seconds). Search for something, like a phrase: You’re so cool, I’m late, or anything you want. You can stick these clips most anywhere, just copy and paste the link or use the social links to post. Our mobile apps (iOS and Android) are designed to share Yarns in Facebook Messenger, find them here and give them a spin.

We process our videos in a variety of different ways. We use artificial intelligence, computer vision, and natural language processing to analyze the audio and video content of the clips. Our database currently has 44,381 distinct words in a vocabulary learned from movie and tv shows (more than the average native speaker)— but not to dissimilar to how many people learn the english language globally.

So if you want to say “hi” with Yarn, there are 518 clips. Or “hello”, a few more with 815. Using our search will drive you so some of these.

In future posts I’ll reveal more statistics and how we do things. Take a look around and let us know what you think.