Introducing….TV Kitchen!

Welcome to the first in an occasional series of updates from TV Kitchen, an open source project that gives you tools to take TV streams and turn them into structured data–such as captions–to power media analysis, fact-checks, investigative research, and more.

Why TV Kitchen?

Local television news still attracts some 20 million nightly viewers, and remains the most trusted source for news. Major media conglomerates, including Sinclair, Tegna, and Nexstar, have been snapping up these local stations, and now control nearly 40 percent. And despite the rise in social media advertising, six publicly held local TV station companies collected $1.2 billion for political advertising during the 2018 elections.

But for researchers and journalists, local TV remains opaque. If you want to analyze local coverage of recent protests for racial justice, the pandemic, or the election, you need to pay an expensive private firm or ignore this major source of information.

What is TV Kitchen?

TV Kitchen is an open source tool to get data out of local TV streams–starting with captions, and in the future political ads, chyrons, talking points, and more. Participants will be able to develop, share, and use free software tools to extract data from TV and share metadata safely with a wider community. The more people who use TV Kitchen, the more plentiful and rich our information and collaborations.

Progress on TV Kitchen

Thanks to generous funding from the Knight Foundation, the TV Kitchen team has spent early 2020 developing the underlying open source code that will power the TV Kitchen community. We’ve reached some milestones:

• BUILDING: Live TV Ingester. The ingester will work with an HDHomeRun to allow users to plug into a local TV stream and collect live TV.
• BUILT: Local Video File Ingester. The ingestion engine takes video streams and converts them into standardized packages of data, for ease of sharing and building tools.
• BUILT: Caption Extractor. The first “appliance” that will be available for the TV Kitchen community will be a caption extractor — you run it against an ingested stream and extract closed captions in real time.
• BUILT: Line Cooks. Passes data to and from appliances to run a particular “recipe.”
• BUILDING: Countertop Coordinator. Loads recipes, creates line cooks to handle them, and shares the resulting data with the TV Kitchen API.
• BUILDING: TV Kitchen API. Receives data from the countertop coordinator and shares it with anybody listening.

Welcome to our partners!

We welcome our first partners on TV Kitchen, who have agreed to take it for a spin and run some experiments.

Center for Cooperative Media at Montclair State University: The mission of the Center for Cooperative Media is to grow and strengthen local journalism, and in doing so serve New Jersey residents. The Center will be using TV Kitchen to collect local television news from seven local stations, for use in analyzing how TV news is covering issues, asking questions such as “Which towns are discussed the most, and by which outlets?” and “Which people are discussed the most?”

Politico: Politico, a national political outlet, will use the TV Kitchen API to pull caption feeds from at least one network during a notable event — such as a presidential debate — and shunt it to a tool that allows its reporters to add commentary to the live transcript.

PolitiFact: The national fact-checking site PolitiFact will use TV Kitchen to capture captions from Sunday talk shows, major news events, or press conferences, to use as a basis to determine what claims need to be fact-checked. PolitiFact may also set up local TV collection spots with members of its nationwide network.

Would you like to partner with TV Kitchen? Contact us at tvkitchen@biffud.com

What’s next?

TV Kitchen is beginning with tools to extract and share captions from TV streams, but it won’t stop there. The future could include new modules that allow community members to do increasingly complex analysis with a rich variety of data: chyrons, political ads via audio fingerprinting, named entity extractors, and more. The community is open, so we welcome developers who want to contribute code and ideas to the benefit of all.

How we got here

In 2016, the TV Kitchen team worked with the Internet Archive on the Political TV Ad Archive, tracking political ads on TV and turning them into data. This fueled all kinds of creative uses, including an online video game, a visual on how CNN, Fox, and MSNBC covered presidential debates, and fact-checks of political ads. TV Kitchen takes this same functionality and invites the public to collaborate.

TV Kitchen is a project of the Bad Idea Factory, a collective of chaotic creatives using technology to make people thinking face emoji. BIF has only existed for a few years, but our members have been developing software for journalists for the better part of a collective century.

We are collaborating with the Media-Data Research Consortium (M-DRC), which is working to make TV news (and other media) accessible for computational analysis by scholars, journalists and others. The Consortium partners with the Internet Archive’s TV News Archive to enhance public benefit access to its ongoing collections of broadcast and cable TV news. Recent M-DRC experiments in understanding how COVID-19 is being covered differently than past epidemics can be seen here, here, and here.