It’s the moment everyone has been waiting for all year: Spotify wrapped time Although it launched later than originally planned, the personalized annual digest is available for subscribers to see where their musical tastes fared during the year. Also available is a custom podcast courtesy of NotebookLM by Googlewith two AI hosts struggling to pronounce any song or artist that isn’t of the Anglo-Saxon variety.
Artificial intelligence is not perfect. Even with the most convincing demos at developer conferences, AI text-to-speech always sounds slightly forced compared to how humans speak. The best example is Spotify’s Wrapped AI podcastwhich resembles Google’s demo earlier this year a Google I/O. The hosts are certainly compelling, but the longer you listen to the synthetic podcast, the more obvious it becomes that you’re listening to a couple of bots summarizing your listening stats.
The AI-produced podcast begins in a standard way. It looks like me and my podcasting partner on an episode of our show. First, the hosts highlight the longest day you’ve listened to Spotify for the year. It was picked up on a particularly emotional day when I was listening to hours of sad music. The AI hosts wondered aloud if it was because he was on a road trip that day. (I wasn’t. I was crying!) This is an opportunity where the AI could have reduced the guesswork. Like a friend who doesn’t think about the impact of his words, AI reminded me of a day of major upset. Rude behavior!
Then AI hosts read your most played artists and songs of the year. I’d still rather hear a real, live human attempt to pronounce the name of my favorite Dutch artist, Joost Klein, than Google’s NotebookLM. His name is pronounced “Yost,” but the podcast refers to him as “juice.” It’s even worse when they try to pronounce the Dutch songs. My most streamed song this year is called “Wachtmuziek” and the Google NotebookLM hosts pronounce it exactly as you might try to read it as a non-Dutch speaker, minus the lucky hawk which is pronounced with the “cht”.
Spanish speakers will also find that Google’s NotebookLM struggles with the Latin language. My publisher shared their custom podcast with me and I cringe every time they try to identify a Bad Bunny song. Don’t get me started on the butchered pronunciation of “ray-gay-ton,” which is no how do you say reggaeton out loud. I would say pronounce the word with the right intonation is the gender The cringe factor is high with this one.
This leads me to wonder: isn’t this something AI should be picking up? Music is classified according to the language in which it is performed. I imagine a bot could be programmed to pull this data and adjust accordingly. To its developers’ credit, Google covered its bases by mentioning that its AI hosts occasionally mispronounce words. But it takes away the AI magic when you realize that the bot has the same flaw of not practicing pronunciation as the rest of us.
Spotify’s Wrapped AI Podcast with Google Skills is available to Free and Premium users in the US, UK, Australia, New Zealand, Canada, Ireland and Sweden for a limited time.
