Earlier this year, Pandora challenged Spotify’s personalization capabilities by using its Music Genome technology to create dozens of customized playlists for its users. Today, it will begin leveraging similar technology to do the same for podcasts. The Podcast Genome Project, as it’s called, is now powering a new recommendation system that will be combined with human editorial oversight to offer content suggestions for Pandora users.
Like the Music Genome — the music information database capable of classifying songs across 450 different attributes — the Podcast Genome Project is also a cataloging system designed to evaluate content. But in this case, its focus is on audio programs.
Says Pandora, the Podcast Genome Project can currently evaluate content across over 1,500 attributes, like MPAA ratings, production style, content type, host profile and more, as well as listener signals, like thumbs, skips, replays and more. It uses machine learning algorithms, natural language processing and collaborative filtering methods to help determine listener preferences.
Also similar to the Music Genome, the Podcast Genome technology is combined with human curation to make its recommendations.The system will do more than suggest shows to try out, Pandora notes. In addition to finding new podcasts, it can also suggest which episodes to listen to at the right time.
The Pandora app will direct users to its recommendations in the app in a variety of ways, primarily within a new “Recommended Podcasts For You” Browse module. In addition, listeners will be able to discover new podcasts throughout the listening experience on Pandora, in the now playing screen, in the search bar, in the podcast backstage passes and in the episode backstage passes, a company spokesperson tells us.
The goal with the new project is to help solve the issues around podcast discovery, which is an increasingly difficult challenge as the genre’s popularity explodes with more podcasters creating shows. However, Pandora points out that the majority of U.S. users aren’t regular podcast listeners.“It might feel like podcasts are ubiquitous, but, eighty-three percent of Americans aren’t yet listening to podcasts on a weekly basis, and a majority of them report that’s because they simply don’t know where to start,” said Roger Lynch, Pandora CEO. “Making podcasts — both individual episodes and series — easy to discover and simple to experience is how we plan to greatly grow podcast listening while simultaneously creating new and more sustainable ways to monetize them,” he said.
Pandora was recently acquired by SiriusXM for $3.5 billion — a deal that will likely bring some of SiriusXM’s exclusive audio programming over to Pandora, increasing its lineup of available audio shows. Technology that’s capable of analyzing these sorts of programs for the purpose of recommendations could be useful there, too, in the future.
The personalized podcast destination is launching into beta today to select Pandora users on iOS and Android devices. It will expand to the wider public in the weeks ahead.
https://ift.tt/2B4yZFq
0 Response to "Pandora brings its Genome technology to podcast recommendations - TechCrunch"
Post a Comment