3 Things: Synthetic Voice for Podcast Ads, User-Curated Podlists, and Immersive Audio Tours

The theme for this week is audio, audio, audio!

Happy Sunday! Each edition of 3 Things will contain a dive into 3 rabbit holes I’ve found myself going down recently. Subscribe to get each week’s edition straight to your inbox and if you enjoy it, please share! This past week, honestly I spent most of my time worrying about the future of the United States… but when I wasn’t doing that or checking my phone every 10 minutes, I was thinking about:

• Synthetic Voice for Podcast Ads

• Playlists for Podcast Discovery

• Immersive Audio Tours

1. Synthetic Voice for Podcast Ads

I’ve been obsessed with podcasts for over a decade. Literally since the iPods began supporting RSS feeds and This American Life became available as a podcast. I’m generally an auditory learner but a much bigger pull for me is the fact that audio is a companion medium that lets me multitask. I can be walking, running, cooking, biking, doing dishes and still be listening. After the launch of Serial in 2014 (which is still the most downloaded podcast of all times having over 340 million times as of 2019) podcast listenership exploded and has continued to steadily rise. Today there are 1.5 million shows up from 550k only 2 years ago and over 68 million people listen to podcasts weekly in the US.

While Serial might be the reason so many people listen to podcasts, they did not originally capture the value they were creating. The way that podcast ads worked historically (and for the most part still work today) are through direct response ad sponsorships where an advertiser pays a flat fee per episode based on the estimated number of listeners (usually $10-30 per thousand listeners) and provides ad copy that the hosts reads and places in the beginning or middle of a static mp3 file. There are a few major flaws here. First, podcasters don’t have accurate listener information as most podcatchers (Apple podcasts, Spotify, Stitcher, Overcast, etc) auto-download shows that a listener is subscribed to and the podcasters only see # of downloads, not # of people who fully listen to an episode, so the data they share with advertisers is inaccurate. Secondly, when an ad read is hard-coded in an episode, there is zero opportunity to personalize ads (based on demographics, location, etc), monetize a back catalog, or cap an advertiser to a certain number of downloads.

Today, some companies are starting to do dynamic ad insertion like RedCircle and even Spotify but there are limitations. Host-read ads are still the gold standard but how many variations of ads is a podcaster willing to record for a single episode? Usually not many which means that you either can’t do host-read ads or you can’t scale personalization. I think there is an interesting opportunity to leverage voice synthesis technology (companies like Lyrebird which as acquired by Descript, Resemble.ai, Vocalid and others are making great strides in this area) to create a truly scalable, personalized, realtime bidding ad platform for podcasts. You could train a model on all of the existing podcast content to create a “voiceprint” of each podcaster. Advertisers can then bid to target anonymized individuals, based on their demographics and divorced from the podcast they’re listening to, on a platform akin to the Google and Facebook ad platforms. When an advertiser wins the bid, the episode is stitched together with the ad read in the host’s “voice” and delivered to the listener. This could be a way to close the monetization gap that exists in the podcast ecosystem.

Curation Platform for Podcast Discovery

To continue on the theme of podcasts, we now turn to the listener side and the biggest pain point we experience today which is discovery. Most listeners learn about new podcasts by word of mouth. People tell their friends, colleagues, and family about shows they like and that remains the #1 discovery mechanism today. In a world where everything has a personalized recommendation algorithm, it’s somewhat archaic that podcasts have basically nothing even remotely close to that.

In addition, podcast discovery is still for the most part tied to an entire feed and not a single episode which seems extremely silly as the majority of podcasts are episodic and not serial. For the podcasts I subscribe to, I rarely listen to all episodes in the feed but pick and choose based off of individual episode topics or guests. There are many people who are taking stabs at solving discovery through interesting means. Companies like Shuffle are using snippets where listeners or podcasters can create a bite-sized clip of an episode by selecting parts of a transcript, add text/images, and share to social media to gain awareness and new listeners. Others are attempting Blinkist-like summaries.

An opportunity I’m interested in would be to create a platform for user-curated playlists of episodes based on certain topics. Think Spotify playlists where users create lists of songs for moods or activities like “work mode”, “workout pump-up”, “Sunday morning” but instead they would be playlists of podcast episodes on things like “best interviews with Elon Musk”, “advice for bootstrapped founders”, or “10 best true crime stories”. Curators build playlists and then share to their audiences on social media which create an inherently social element and the chance at virality. They build status and clout and can showcase their thought leadership (plus gain followers) on certain topics by curating lists and they help their followers discover new episodes and shows. Users can download the playlist to their podcast app of choice as its own feed so it doesn’t require changing apps or adding additional friction for listeners. It also solves the problem of having to subscribe to entire podcast feeds.

Immersive Audio Tours

Have you ever paid for a guided tour on vacation? The experience is completely dependent on how good your guide is. A great guide can make even a not-so-interesting place come alive with history, energy, and excitement and a bad guide can ruin even the most majestic of sites. When you’re traveling, you have limited time, budget, and information and often just get guide recommendations from your hotel or the site itself (people are always standing just outside of major landmarks trying to sell you their guiding services). Other times you might not be at a noteworthy site but would still be interested in hearing about the history of the area. I’m a big hiker and I often find myself searching for podcasts about the place I’m traveling and listening to them while I’m hiking.

On a trip to Bolivia last year, I went on a 16 mile hike by myself up and over a large mountain and had downloaded a podcast about the Incas. It was horribly produced and, to be honest, pretty boring and I found my mind constantly wandering. I started thinking about some of the best storytellers I knew and the best guides I’ve had in the past and yearned for a podcast that was like that. I started noodling on this idea and the way I’d build it. I’d start by finding the best guides and historians for major sites and cities and producing high-quality immersive audio tours. Think Masterclass or Knowable quality content and leveraging your phone’s GPS so it can cue different parts of the audio tour based on your location.

There was an attempt at something similar a few years back by Andrew Mason the founder of Groupon, called Detour (which I personally loved) but it was never able to gain traction and was sold to Bose in 2018 and shut down. There are few things I would do differently. First, I’d sell through all of the companies that touch a traveler (and not direct to consumer) such as booking sites, AirBnB, luxury travel companies, active travel companies like Backroads, hotels, etc and would put an emphasis on B2B2C where the buyer is the company not the traveler and they give it away for free. Second, I wouldn’t try to create the content wholly from scratch but leverage the people, stories, and scripts that are the best for each place and use audio techniques to make the experience feel high quality and immersive. Finally, while I think that AR is very cool and we will see a lot more of it over the next decade, I wouldn’t overcomplicate this and focus on the audio experience alone, letting the user focus on being in the place they’re visiting not staring at their phone.

That’s all for today! It’s been a rough week but hopefully everyone can now relax a bit. If you have thoughts, comments, or want to get in touch, find me on Twitter at @ezelby and if you enjoyed this, please share with a friend or two!

~ Elaine