Subscribe
Sign in
Start new thread
p/very-disco
Everything ever mentioned in your favourite podcasts
Visit Product
Desi Cochrane
Very Disco — Everything ever mentioned in your favourite podcasts
Featured
29
•
Meet Very Disco — the “IMDb for podcasts”. Discover, listen to, and save every book, movie, TV show, and person ever mentioned in a podcast.
Replies
Desi Cochrane
Maker
📌
Hi PH community, I'm Desi, and I’m happy to introduce you to Very Disco — “Everything ever mentioned in your favorite podcasts” Very Disco is the IMDb of podcast content. We identify every topic in every podcast episode: every book, person, TV show, movie, and more. So next time you’re listening to a podcast episode, and an interesting book is mentioned, you can: 1) see which book it is, 2) see every other time the podcast has mentioned it, and 3) find every *other* podcast talking about it too! No more detective work is needed. Once you’ve found it, simply add the book to your Very Disco to-do list and add the episodes to your queue. So whether you want to find every time Paul Graham has been mentioned across all 700 episodes of the Tim Ferriss Show (it’s been 43 episodes so far!), or the 21 episodes where Lex Fridman brings up 2001: A Space Odyssey, or you simply want to discover which topics are discussed most often in Huberman Lab: Very Disco has you covered! What else can you do? 1. Listen to any episode. 2. Read and search episode transcripts. 3. Save, rate, and review any book, movie, episode, or TV show. 4. Add episodes to your queue. 5. Switch between light and dark mode based on your preference. And we’re only just getting started. We will soon be adding more social features, more topics, alerts, and ever-improving AI models that make the magic happen. We are currently featuring seven podcasts, and we will be adding more and more over the coming weeks and months. Let us know in the comments which podcast you'd like us to include next — we’ll add the most-voted podcasts as soon as possible. It's free to use, so give it a try. I'm here for any questions. Hope you enjoy it!
Desi Cochrane
Maker
@mohammed_mustafa_jafer Hey Mohammed, I'm glad you like what we're building! We are super proud of the pipeline we're building for identifying entities. We first ingest the audio and extract any metadata in the audio file itself. Then, we transcribe the audio and identify different speakers. Next, using that context (the metadata, text, audio, and podcast) we have trained embeddings (and a few other tricks) to map speakers to actual people. Next, we have a model that uses all that context, plus extracted language features, we can parse the transcript to identify "topics" (person, book, movie, company, etc). Then, finally, we use all that context combined, with some Bayesian priors, to map these topics to actual entities in our knowledge base. We are currently finishing up a *big* training loop which looks to be a *dramatic* improvement in precision and recall. Plus, we are currently ingesting up a big knowledge base dataset from Wikimedia (and a few others). Armed with all that, we will be rolling this out to the next 50, 500, and 5,000 podcasts over the coming weeks and months - so stay tuned!
Bill Franklin
💡 Bright idea
This is awesome, it's like a Mention / search engine for podcast content. Interested to see where you take this! I'd happily pay for advanced search features, I'm guessing a ton of companies would also be interested in knowing when they or their domain get mentioned on a podcast. Congrats on the launch!
Desi Cochrane
Maker
@billfranklinuk Thanks Bill! Yes, we have some pretty neat alert functionality coming up next on the roadmap!
Simon
💡 Bright idea
Great product y'all! I often listen to podcasts that mention other podcast episodes and/or invite hosts from other podcasts so it's great to see a tool that makes this linking and discovery easier. I'd also love to be able to browse specific topics, e.g. "ADHD" and listen to the most popular and best-rated postcasts in that category. Looks like that's your vision here so I'm rooting for you!
Desi Cochrane
Maker
@the21nd Thanks Simon! Great callout re adding more topics. We are training some models at the moment to do just this and the results are looking super promising.
Klajdi Turlla
💡 Bright idea
Congrats on the launch, super cool. Just finding book recommendations from a particular guest is super valuable
Desi Cochrane
Maker
@klajdi_kl Thanks Klajdi! So many of my favorite books came from an obscure mention in a podcast episode. We're gonna spice up recommendations a lot over the coming weeks.
Jakub Žitný
Hey guys, super cool idea "imdb for podcasts". This might finally get me out of my Spotify comfort to use a dedicated podcast app. Good luck!
Desi Cochrane
Maker
@jakubzitny Thanks Jakub, hope you like it! If your fav podcast isn't there let us know.
Salih Mujcic
Awesome tool! I'm always scrambling to take notes of topics, frameworks, people, products etc etc mentioned. This is fantastic!
Desi Cochrane
Maker
@salih_mujcic Yes Salih I'm exactly the same! I can never get through any podcast episode without adding several papers, books, and movies to my to-do list.
Juxhin
Excellent work! Really intuitive and helpful, big kudos to you and the team. See some great potential for this as my dedicated podcast app.
Desi Cochrane
Maker
@juxhindb Thanks Juxhin! That's what we like to hear!
Darren Baldwin
Sick idea - definitely going to give this a try!
Desi Cochrane
Maker
@darrenbaldwin03 Thanks Darren, enjoy!
Bhanu Gogia
Super cool & useful product @desicochrane, congrats on the launch!
Desi Cochrane
Maker
@bhanu_gogia4 Thanks Bhanu! Glad you like it
Shahmir Faisal
Congrats on the launch Desi!
Desi Cochrane
Maker
@shahmir_faisal Thanks a lot Shahmir!
Val
I love this. Had this idea years ago so I’m glad to see this exists now!
David Kermaani
As someone in the podcasting space, I find this interesting. Who is your target market?
Nick Barth
Maker
@metadavid Yeah, it's a great question with a wide open answer. We love the idea of becoming a Goodreads/Letterboxd for podcasts, a social-focused podcastDB. However, our transcription & entity linking & tagging opens the possibilities for us to appeal directly to creators, we could come up with a pretty gnarly set of creator tools based on this tech. Our data opens us up to being valuable as an API, informing companies of podcast mentions, helping advertisers place ads based on content. We have a lot of possibilities, and part of our goal with this launch is to gauge interest and allocate our resources effectively. As someone in the podcasting space, your opinion is mega-valued, what do you think?
sree sreenivasan
Great idea. Looking forward to exploring more. Two podcasts I could use this for aren't in the index yet: "Empire Podcast" https://linktr.ee/empirepoduk and "No One Listens to Paula Poundstone" https://nobodylistenstopaulapoun... ...
Megan Lieu
Podcast enthusiasts unite! Definitely going to use this to skip to the good parts of my favorite pods