As a huge NotebookLM fan, when Spotify and Google announced a Wrapped AI Podcast last December, saying I was excited is probably an understatement. I was thrilled, and the first thing I did was rush to check it out...only to realize it wasn't available in my region. Unfortunately, the podcast was only available to Free and Premium Spotify users located across the US, UK, Australia, New Zealand, Canada, Ireland, and Sweden.
And given I live in none of these countries, I was left out of the fun. I tried to use a VPN and every other workaround I could think of, but nothing really worked. At that time, I gave up and silently watched everyone online listen to their podcasts. Spotify Wrapped 2025 is not too far away, and I'm still not over being left out. A couple of days ago, I had what I'd consider an absolute genius idea: why not create my own (and possibly better) version of the Wrapped AI Podcast? Of course, that's exactly what I did.
Here’s how I set it all up, step by step
It took less than five minutes to get started
When I do these experiments, I usually have a pretty good idea of how I'll set it up and know what to expect. Other than the fact that I hadn't heard the original Wrapped AI Podcast myself, this experiment wasn't too different. The main (and pretty much only) thing I needed for this was my listening history, which included a record of the songs, albums, and artists I'd spent the past year listening to. Now, while Spotify does let you download your listening history, which includes the name of the track you listened to, how many milliseconds you spent listening to it, and more, it takes around five days to receive the files. I personally didn't want to wait that long once this idea had popped into my head.
Fortunately, I had connected my Spotify to Last.fm in November 2024 (a month before Wrapped 2025 came out). So, I decided to use my Last.fm data instead. It already had a detailed record of my listening habits over the past ten months and had everything I needed to feed into NotebookLM. The next challenge was figuring out how to download the Last.fm data. Thankfully, there are a bunch of tools that let you download your Last.fm listening history. I used a website called BenjaminBenBen.
The process of exporting the data was extremely simple. All I needed to do was enter my Last.fm username, and hit Fetch. After a couple of minutes, I had a clean CSV file with all my listening data divided into four columns: Artist, Album, Track, and Timestamp.
The next step was adding this data to a NotebookLM notebook. Since NotebookLM doesn't currently accept CSV, XML, or JSON files as sources, I decided to convert it to a compatible format: PDF. I then created a new notebook in NotebookLM and simply added the PDF as a source.
Now, here's where the fun part begins: actually generating the Wrapped-style Audio Overview. Instead of simply generating an Audio Overview as-is, I decided to add a custom prompt just to make sure it was as close to the real Spotify Wrapped podcast experience as possible. Here's the exact prompt I went with:
You’re the two AI hosts of Mahnoor’s personalized Wrapped-style podcast. Using the uploaded PDF of her listening history (including artists, albums, tracks, and timestamps from November 2024 to October 2025), create a lively, conversational audio recap of her year in music.
Speak like Spotify’s Wrapped AI podcast hosts — upbeat, funny, and personal, but also insightful. Explore her listening patterns across the year: which artists, genres, or albums she returned to the most, when her taste shifted, and any standout listening moments or trends.
Include natural banter between the two of you (like old friends joking and reflecting together), weaving in data-based insights such as:
– Her most-played artists or songs.
– Notable genre or mood phases over time.
– Seasonal changes in listening (e.g., mellow spring, energetic summer).
– Surprising discoveries or repeat obsessions.
End with a short wrap-up celebrating how her taste evolved throughout the year.
According to a lot of people I asked, the original Spotify Wrapped AI podcast was relatively short — around three to five minutes. Since NotebookLM lets you customize the length of Audio Overviews and choose whether you'd like a Short, Default, or Longer podcast, I decided to generate two versions: a Short and Default one. Then, I waited for NotebookLM to whip the Audio Overviews up.
So, how did my makeshift Wrapped AI podcast turn out?
Finally, no more FOMO
Right off the bat, I thought the podcast NotebookLM generated was quite impressive. I had a ton of fun listening to it, and it was interesting to see how NotebookLM interpreted my listening habits. It managed to highlight my top artists (I confirmed this by comparing it with my Last.fm stats), my most-played tracks, and even some patterns I hadn’t consciously noticed, like how I listened to certain music more in some months than others.
I asked some of my colleagues and friends who were in the regions where the official Wrapped AI Podcast was available to listen to theirs, just so I could compare my makeshift version with the original. Since it was available for only a limited time, none of the people I asked still had access to it. So, I shared my version with them and asked for their feedback. When I shared the one I had generated using the Default length setting (which ended up being around 10 minutes long), one of my colleagues thought the podcast was way too long and that the hosts were trying too hard to read into the patterns, making some parts feel a bit overanalyzed.
That’s when I generated the Shorter version, and another colleague noted that they thought the "analysis is a little more robust than what I remember from mine last year." They also explained that they "don't recall them talking so much about what I listened to during any given time of the year like they did." While one colleague thought Spotify's AI podcast was a bit more wholesome, another remembered the original being more bland.
The next thing I decided to do was check YouTube and see if anyone had shared their official Wrapped AI Podcast. I wanted to get a better sense of the style and pacing so I could compare it more directly with my NotebookLM version.
Once I found one, the very first thing I noticed was that the official podcast referenced official Spotify data like "top 1% of listeners," which my version didn’t have. Of course, that isn’t surprising, given that my version was entirely based on Last.fm data rather than Spotify’s internal metrics.
My version didn't include how many new artists I had explored, how many times I played my top track, how many unique songs I heard, or which day of the year I listened to the most music. My own version was definitely a lot more detailed and went deeper into my statistics. For instance, my third-top artist was Louis Tomlinson, and I listened to his Live album more than his Studio album.
NotebookLM's hosts even pointed that out! It also highlighted certain months when I'd listen to artists I barely touched the rest of the month, like the sudden surge in Ariana Grande that happened in July, which I hadn’t consciously realized until I heard it discussed in the podcast.
Pacing-wise, I personally didn’t notice much difference between the original and mine. The hosts sounded similar, and both podcasts moved at a relaxed, conversational speed. They threw in a couple of jokes in between, which is on-brand for NotebookLM's audio overviews. Of course, this wasn’t surprising — the Spotify podcast was generated by NotebookLM after all.
I'm going to do this every year
I do a lot of these wild NotebookLM experiments, and this might just be my favorite one so far. Given Audio Overviews is my favorite NotebookLM feature, I felt majorly left out when I didn't get to experience a Wrapped-style overview of my own listening habits.
So, I'm glad I decided to do this. It also felt like an improved version of the original, so even if you did have access to Spotify’s Wrapped AI Podcast, I'd recommend trying this out. It only took me around five minutes to set everything up, and I had a blast listening to a podcast all about my own year in music.