1 month ago
22
An AI quality app called Particle, from former Twitter engineers, tin present support up with quality breaking connected podcasts arsenic good arsenic quality published connected the web.
Just up of its caller Android release, Particle has introduced a diagnostic called Podcast Clips, which finds the astir absorbing and applicable moments crossed galore antithetic types of podcasts, and past includes those clips alongside the related quality stories successful its feed.
So alternatively of listening to a lengthy podcast conscionable to drawback the 45 seconds of absorbing comments, you tin play backmost the clip arsenic you’re speechmaking the quality connected Particle. You besides person the enactment of speechmaking the transcript of the clip instead, arsenic the words are highlighted arsenic they’re spoken.
Image Credits:Particle“We’ve done that fundamentally for immoderate quality communicative — if determination is simply a podcast that is talking astir it, oregon applicable astatine all, we’ve got each those clips,” Particle CEO Sara Beykpour, previously the Senior Director of Product Management astatine Twitter, told TechCrunch. “It’s a truly chill way, erstwhile you’re speechmaking a communicative oregon learning astir a story, to get a enactment of what are radical saying astir this? What’s the commentary?”
The summation acknowledges a displacement successful the quality ecosystem that’s been underway for years. Not lone are much radical getting their quality from podcasts and trusting them arsenic reliable sources, but the mean is besides becoming a destination for breaking quality and large announcements from nationalist figures.
Tech CEOs, successful particular, are present seeking retired affable podcast hosts to aerial their talking points alternatively of trying to enactment with accepted media, arsenic Bloomberg reported successful 2024.
That makes paying attraction to podcasts adjacent much captious if you privation to support up with news.
Beykpour says Particle uses embedding models to recognize erstwhile podcasts subordinate to a fixed quality story. These models are provided by the aforesaid companies that supply LLM models, but they’re not generative AI technologies, she explains.
“We usage vector embeddings to recognize that these antithetic parts of the podcasts are related to these antithetic stories,” Beykpour notes. “A azygous podcast mightiness screen 10 oregon 20 stories, truthful we usage AI to recognize that. We besides usage AI to bash immoderate of the logic astir clipping, and knowing erstwhile to commencement a clip and extremity a clip.”
Image Credits:ParticleThe institution leverages exertion from ElevenLabs for transcription. However, immoderate of the exertion that identifies wherever precisely to clip the audio is portion of Particle’s concealed sauce.
The thought to pat into podcasts to amended recognize the commentary astir quality is besides thing newsrooms are taking a person look astatine these days. As Nieman Lab reported this month, The New York Times has been utilizing a customized AI instrumentality that employs LLMs to transcribe and summarize caller episodes of dozens of right-wing and much blimpish podcasts to amended recognize what influencers connected that broadside are saying astir the news.
Particle’s Podcasts Clips diagnostic isn’t lone tied to quality stories. Because the app already understands antithetic entities — similar people, places oregon things — you tin spell to the leafage for a notable figure, specified arsenic OpenAI CEO Sam Altman, to spot each of his appearances connected podcasts arranged arsenic a feed.
Image Credits:ParticleParticle has been engaged gathering different features, arsenic well. The institution has made its archetypal effort astatine monetization with Particle+, an optional $2.99/month subscription (or $29.99/year) that lets you entree premium features. These see the quality to: usage earthy connection to person the quality summarized successful a benignant you prefer; prime from antithetic voices erstwhile utilizing the personalized audio feed; “Listen to the News”; unlimited crossword puzzles; enactment for backstage questions with its AI chatbot, and more.
Image Credits:ParticleThe Android merchandise besides brings a mates of different notable changes. The browse tab present includes timely stories, similar the 2026 Winter Olympics, successful summation to emblematic sections similar politics, tech, oregon entertainment. Plus, erstwhile you pat connected an entity, you’ll spot a caller leafage with the definition, stories, articles, related entities, and related topics.
Image Credits:ParticleParticle isn’t sharing information astir idiosyncratic enactment oregon conversion rates, but Beykpour did constituent to the app’s planetary audience, pre-Android. On a play basis, 55% of Particle’s users are extracurricular the U.S., with India (15%) its biggest marketplace aft the U.S.
English (US) ·