Could podcast transcription help with moderation?

·3-min read
A new function from Spotify will let listeners read the content of certain podcasts on their smartphone.

Giants of the tech world are doubling down on innovations to stand out in the highly competitive podcast market. To this end, Spotify is working on bringing auto-transcription to its audio formats, an initiative that will no-doubt boost inclusivity, but which could also help tackle a major problem faced by podcast hosting platforms: moderation.

Subtitled podcasts are coming soon to Spotify. The firm has announced the forthcoming arrival of automatic transcription to certain podcasts on the platform in the coming weeks. This function will allow listeners to read the content of a podcast on their smartphone, either while they listen or not. "Users will also be able to quickly scroll and navigate through specific episodes and click on any written paragraph to start playing [the podcast] from that point," explains Spotify in a statement.

Although only a selection of original and exclusive productions will be getting auto-transcription for now, the Swedish streaming giant hopes to make this function available for all podcasts on the site. This would once again help Spotify stand apart from Apple and Google, its main rivals in the audio market, although the online search giant already dabbled with transcribing podcasts in 2019, to boost search optimization.

To moderate or not? That is the question

Whether for improving search indexing or for inclusive accessibility, podcast auto-transcription opens up many perspectives for the platforms hosting them. But could it also help improve podcast moderation? While tech giants haven't yet spoken out on the subject, the question of moderating audio content is at the heart of their concerns.

Especially for Spotify. The firm recently came under fire following controversial claims by podcast host Joe Rogan about covid-19 vaccines. "I'm not an anti-vax person. I said I believe they're safe, and I encourage many people to take them," he said at the end of April in an episode of his "The Joe Rogan Experience" podcast. "I just said that if you're a young, healthy person that you don't need it." Words considered irresponsible and dangerous for many critics in the US, including the specialist website Media Matters for America.

Spotify CEO, Daniel Ek takes a different view. When questioned by Bloomberg about the words of the platform's star podcast host, the CEO avoided the question by pointing to the fact that Joe Rogan is one of "eight million creators." He continued: "We have a content policy, and we do remove pieces that violate it." And while the streaming site has censored "The Joe Rogan Experience" in the past, it seems somewhat reluctant to do so -- perhaps because the podcast regularly tops the list of most-listened-to audio content in the US, and even exceeded Spotify's own expectations in terms of new users and engagement during the first quarter of the year.

A colossal task

But the issue of podcast moderation isn't unique to Spotify and "The Joe Rogan Experience" -- it concerns the whole world of podcasts. The Media Matters for America website reported in October that podcasts with QAnon content or linked to other extreme-right movements were multiplying on hosting platforms like Apple Podcasts or Google Podcasts. And that's still the case today.

Automatic transcription of podcast content could, theoretically, contribute to stamping out the problem. But it's far from an easy task, according to Mike Kadin, founder of the specialist podcast startup RedCircle. "It's quite hard to do it at scale," he told The Verge. "We would have to transcribe everything, maybe, and apply some automated filters to look at everything. A: that's expensive, and B: even if we could get everything in text, I don't think a computer can understand the nuance of some of these issues."

Caroline Drzewinski