Day: 31 December 2021

In late December, for a few years now, I have tweeted out a big list of albums I enjoyed. On Thursday, I posted my picks for 2021, some of which were likely not a surprise for anyone who follows me on Last.fm. This spring, I reactivated my account there and began scrobbling again after years away in the pursuit of better music recommendations. I am not sure it is working, but here is what I have found so far.

Apple Music is a remarkable deal for me: spending ten bucks a month gives me access to almost any record I can think of, often in CD quality or better. There are radio features I do not use and music videos I rarely watch, but the main attraction is its vast library of music. Yet, with all that selection, I still find new music the old-fashioned way: I follow reviewers with similar tastes, read music blogs, and ask people I know. Even though Apple Music knows nearly everything I listen to, it does a poor job of helping me find something new.

Here is what I mean: there are five playlists generated for me by Apple Music every week. Some of these mixes are built mostly or entirely from songs it knows I already like, and that is fine. But the “New Music Mix” is pitched as a way to “discover new music from artists we think you’ll like”. That implies to me that it should be surfacing things I have not listened to before. It does not do a very good job of that. Every week, one-third to one-half of this playlist is comprised of songs from new albums I have already heard in full. Often, it will also surface newly-issued singles and reissued records — again, things that I have listened to.

When I scroll down to the “New Releases” section on the “For You” page, it is an even sadder story. Perhaps I have this all wrong, but this seems to me like it should be where I learn about new albums from artists I already listen to. I can remember just one time since Apple Music launched when this section matched my expectations for it. At all other times, it shows weeks-old records I have not played from artists I have not heard of. And they just sit there for weeks, unplayed, until another set of similarly-confusing picks is displayed. Have I got the concept of “New Releases” completely wrong?

Shallowest of all are the “Similar Artists” recommendations on every artist’s page. It tends to prioritize proximity to the selected artist, so it often shows side projects and solo acts. For example, according to Apple Music, artists similar to Soundgarden include Chris Cornell — who was Soundgarden’s lead singer — and Temple of the Dog — one of his side projects — and Audioslave — another Cornell project. It also suggests Alice in Chains, Stone Temple Pilots, and Pearl Jam, three other bands with similar tonal qualities. How many listeners of Soundgarden are there in 2021 who do not know about any of these other bands and projects? I would wager it is a tiny number given Soundgarden’s fame and fanbase. I suppose there are some people who are not fans, per se, and would appreciate these recommendations. But why is Apple Music showing me those artists when I have listened to them all in Apple Music?

In fairness, the artist pages are distinct from the “For You” section of the app. Yet, surely the entire service should be tailored for me. Otherwise, what is the purpose of the algorithmic backend?

You may rightfully ask why I have not stopped using Apple Music and switched to, for example, Spotify, which has far better recommendations. The answer is because I have an anachronistic setup of mostly local music that I would like to keep syncing to my iPhone, and I still do not trust any of the matching or cloud syncing features to do that job for me, including Apple’s.

So: Last.fm. There are a few things I like about it. First, it seems to take into account my entire listening history, though it does give greater weight to recency and frequency. Second, it shows me why it is recommending a particular artist or album. Something as simple as that helps me contextualize a recommendation. Third, its suggestions are a blend of artists I am familiar with in passing and those that I have never heard of.

Most importantly, it feels free of artificial limitations. Apple Music only shows a maximum of eight similar artists on my iPhone, but there are pages of recommendations on Last.fm. Echo and the Bunnymen has twenty-five pages with ten artists each. I can go back and see my entire listening history since I started my account there. Why can I only see the last forty things I listened to on Apple Music?

There are so many things Apple could learn from Last.fm’s recommendation approach, and I wish it would. Right now, its approach is somewhere between inconsequential and unhelpful. It does not have to be this way, and it should not be this way.

Maybe part of my appreciation comes from my nostalgia for the mid-2000s internet era. They are memories of shiny, colourful logos, wet floors everywhere, and new social networks for every conceivable interest. These websites encouraged centralization and many were ultimately destructive to privacy, but there were also gems like Last.fm. It was built around a simple premise: track your music listening history for better recommendations.

It still feels like an artifact of a simpler era. While Apple is busy rebuilding Music in MacOS so it feels less like a weighty mess, Last.fm still feels like a breath of fresher air. I am not calling it lightweight — it is still a web app, so that would be ridiculous — but it does not feel as ponderous as Apple’s attempts. I wish Apple could capture a bit of that magic, if only because Music is still used every day on all of my devices.

In the meantime, I will keep tracking my library with Last.fm. It feels a little quaint, a little cute, but I like it. On my Macs, I use NepTunes; on my iPhone, I use Soor. Both are very good.

Tom Parsons of What Hi-Fi? recently interviewed Apple’s VP of acoustics Gary Geaves, and Eric Treski, who works on AirPods marketing. This part seems worth thinking more about:

This is where Adaptive EQ, which was first introduced with the AirPods Pro, comes in: “we’ve added an inward-facing microphone”, says Geaves, “which continuously monitors what’s being played by the speaker and tunes the bass and, to some extent, midrange frequencies as well, to deliver a really consistent frequency response regardless of the level of fit that each person gets”. The idea is that everyone hears the music the same way, and the way the artist intended.

Geaves’ response has echoes of computational photography about it. When asked to clarify how Apple could possibly know what the artist’s intent could be, Geaves says that it is a mix of analytics and human adjustment. I still get the feeling that we cannot really know — but that it is also true of audio products generally. How do any of us know whether the speakers in our headphones or home audio setup are fairly representing what we are listening to?

Parsons presses the two Apple representatives on new stuff released this year, like the third-generation AirPods and spatial audio. But it is when asked about lossless audio that Geaves gives the most intriguing answer:

“Obviously the wireless technology is critical for the content delivery that you talk about”, he says, “but also things like the amount of latency you get when you move your head, and if that’s too long, between you moving your head and the sound changing or remaining static, it will make you feel quite ill, so we have to concentrate very hard on squeezing the most that we can out of the Bluetooth technology, and there’s a number of tricks we can play to maximise or get around some of the limits of Bluetooth. But it’s fair to say that we would like more bandwidth and… I’ll stop right there. We would like more bandwidth”, he smiles.

Given that AirPods Max and Apple Music’s lossless audio option were announced within six months of each other, yet were incompatible for bandwidth reasons, it seemed like something had to give. It felt like a plot hole in both products’ respective stories.