Showing posts with label meyer. Show all posts
Showing posts with label meyer. Show all posts

Friday, February 28, 2025

Muzak Is Back


Seven years ago I started buying muzak. That's right you heard me, it's on my playlist. And that's right, you too can buy muzak. I learned a long time ago to stop questioning my own behavior and just let it happen - it all makes sense if you wait long enough.

It all started while watching a Japanese television show about the cultural practices around sleep in Japan. (If you fall asleep in class as a college student, it's not seen as disrespectful, because just being in the same room shows you care.) The background music caught my ear, and I looked further into the description for the video, and found the source (go figure, what a crazy idea). 

Then I said to myself - can I buy this? Is muzak only for sale to Japanese television show producers and discount furniture store managers? Yes! No! You can totally buy muzak. In fact, it's easier than buying regular music these days (probably because nobody buys music these days). I found the track I was looking for, even looked up the artists involved and bought a couple more. I left the experience thinking how strange it is to be a human, urges and decisions come from seemingly nowhere, make you do things, but at the same time there's a man behind the curtain, watching the one who does things, bewildered. What is wrong with me? Who on earth is buying muzak, on purpose, like to listen to on purpose? 

Years later, two weeks ago, I sit in a cafe listening to laid back acoustic covers of pop music from the 60's - 90's coming from their streaming platform's playlist. After one too many of these cover songs, I get a chill down my spine - something weird about this, the entropy-measuring machine in my head is confused. I approach the business owner and ask where these songs are coming from. He tells me Spotify, but I already know that. I ask again, are these songs AI? All of them are AI, aren't they? I saw the flash in his eyes. Prompt engineering has been a thing for a while now; people are familiar with it. So this business owner imagines out loud, "young asian with thick rimmed glasses acoustic guitar for low-key cafe background music", like that? Just like that, I said. And he looked at the device controlling the music, and he scrolled through the artists, and he looked at the prompt he originally input. He scrolled down, then back up, then back down again, his eyes feverish and darting. 

The following week, I'm eating lunch in a grand banquet hall for a work conference in Atlantic City, laid back music softly playing in the background. But then it happens - it's Black Sabbath's Ironman, as a bossa nova cover. Now I am no music expert, maybe an aficionado, but by no means a music trivia master. But I will tell you this - if you've ever heard a bossa nova cover of Ironman, you would fucking know. (The Cardigans is as close as you'll get by the way). That's it. I put down my fork, and with a mouthful of food I walk across the banquet hall of 1,000 people and up to the DJ. He's behind a table, manning his decks and his sound equipment, minding his own business. I start off politely, but then I can't hold back - what the hell is this? There's no way this is real. He's confused. I say look at the playlist, you recognize any of these artists? He says who the hell are you. I say look at the playlist. He does. He's still confused. He looks back at me and asks again who I am. But that's not important, I'm not important. What's important is that he - who apparently owns the audio-visual company serving this conference, had no idea this was happening. He just types "bossa nova brunch hour", pushes the button and sits down for the afternoon. I don't blame him, to be clear, although I'm a bit more upset about the guy running an A-V company than the one running a cafe.  

Maybe you run a business. Maybe you like to have music in the background for your customers. Not like the kind of music that makes you wanna cry, or fight, or dance or sing. The kind of music that makes you more relaxed, maybe more focused, but that you don't really notice. Some might call this functional music, but I haven't heard the term used, I'm just making this up. 

Then I realized there's already a word for this; it's called muzak. Start paying attention and you'll hear it too. You can tell because entropy. (See below for links to older posts about how entropy works in music, and information in general.) Entropy is an important concept in the world of artificial intelligence, and once you know what to look for, it helps you understand how it works, and how it doesn't.

Let's make this point real quick - Spotify does not pay musicians anymore, because they don't have to. They made a music generator that's been trained on real musicians' music, and plays that music back again, but in a way that's different enough to be considered Fair Use (under copyright law), and despite that fact that they stole the training music in the first place. You load a prompt, Japanese television show background music for example, and sit back and let the computer make music for you in real time. Guess who gets paid - not the original musicians, not the computer making the music in real time. Spotify gets paid; they get paid by advertisers, and by business owners who want ad-free music. And they pay nobody in return. They have effectively removed musicians from the equation. 

And we have effectively moved right past muzak, and into the low-entropy world of cold, unemotional non-music. This essay by music blogger Ted Goiai does a great job explaining what's happened to the music industry over the past years and even decades, as it becomes more and more boring, less and less able to create something new, or to take risks. Everything becomes plain vanilla, everything becomes a potato. No parsnips, no yams, no yucca, no celery root, no daikon, no turnips, no rutabaga. Just potatoes, because they are the most in-between, the most representative, and ultimately the most boring. The way artificial generation works is ultimately ruled by thermodynamics, and so the output can never have more entropy than the input. In other words, the music being produced can never be more interesting than what it's copying.

Should I be making a plea for people to support their muzak-making co-species? Perhaps. Or perhaps muzak was the first vulture to circle this industry. I'd rather make a plea for business owners who want music to play at their place of business to reach out to local radio station DJs who know where to find real, new music. There is new music being made every day by real humans. It's new, it's real, and it's good, but you would never know, because the consolidation of the music industry, and unfortunately every creative industry, makes everything look like everything else until there's only one song left (probably a Beatles song). If you run a business, you don't have time to find these musicians and their music. But local radio station DJs do, and they can help you populate your playlist. 

Two weeks after the epiphany at the cafe, the owner is back to his old playlist, which is about 500 songs by artists you know and love. No covers, no computers, just music. But two weeks after that, he's back to the muzak (auto-generated aka AI music). We can only listen to Nirvana's cover of The Man Who Sold the World so many times, you know? And the novelty has yet to wear of for the auto-generated, author-vacant, soft-cover version of that song where the auto-translated lyrics screw up Kurt Cobain's mumbled words and repeat them like they're totally normal (the way your GPS has funny ways of pronouncing some of the streets in your neighborhood).

Somewhere between the two stands the music industry, like "Confused John Travolta" in Pulp Fiction.


Post Script from the Music Writer Ted Gioia:
"In fact, nothing is less interesting to music executives than a completely radical new kind of music. Who can blame them? The radio stations will only play songs that fit the dominant formulas, which haven’t changed much in decades. That’s even more true for the algorithms curating so much of our new music—the algorithms are designed to be feedback loops, ensuring that the promoted new songs are virtually identical to your favorite old songs. Anything that genuinely breaks the mold is excluded from consideration almost as a rule. That’s actually how the current system has been designed to work." -Is Old Music Killing New Music? Jan 2022 [link]


Further Reading on How Entropy Works in Music, and Information in General:
Entropy Engineering and Social Syncopation, 2025
Your Brain is a Prediction Machine, 2021
Culture as Learned Probability System, 2012
On Free Will and Risk, 2012
Nature Nature Everywhere, 2017

Friday, January 3, 2025

Entropy Engineering and Social Syncopation


Singing researchers find cross-cultural patterns in music and language
May 2024, phys.org

An international team of researchers recorded themselves performing traditional music and speaking in their native language. In all 50+ languages, the rhythms of songs and instrumental melodies were slower than those of speech, while the pitches were higher and more stable.

Speculating on underlying reasons for the cross-cultural similarities, Savage suggests songs are more predictably regular than speech because they are used to facilitate synchronization and social bonding.

"Slow, regular, predictable melodies make it easier for us to sing together in large groups," he says. "We're trying to shed light on the cultural and biological evolution of two systems that make us human: music and language."

via Max Planck Institute for Psycholinguistics in Nijmegen: Yuto Ozaki et al, Globally, songs and instrumental melodies are slower, higher, and use more stable pitches than speech: A Registered Report, Science Advances (2024). DOI: 10.1126/sciadv.adm9797. 



Neuroplasticity study shows how singing rehabilitates speech production in post-stroke aphasia
May 2024, phys.org

According to the findings, singing repairs the structural language network of the brain. The language network processes language and speech in our brain. In patients with aphasia, the network has been damaged.

"For the first time, our findings demonstrate that the rehabilitation of patients with aphasia through singing is based on neuroplasticity changes"

via University of Helsinki: Aleksi J. Sihvonen et al, Structural Neuroplasticity Effects of Singing in Chronic Aphasia, eNeuro (2024). DOI: 10.1523/ENEURO.0408-23.2024


Study shows the power of social connections to predict hit songs
Jun 2024, phys.org

The team analyzed data from last.fm, analyzing 2.7 million users, 10 million songs, and 300 million plays, along with networks of mapping friendships and another capturing influence dynamics (who listens to a song and who follows suit). 

Examining the first 200 plays of a new song, they predicted its chances of becoming a hit.

The researchers improved the precision of predicting hit songs from 14% to 21%.

Existing models often focus on artist fame and listening metrics, but the CSH study highlights the overlooked social aspect - musical homophily, which is the tendency for friends to listen to similar music.

via Complexity Science Hub Vienna: Niklas Reisz et al, Quantifying the impact of homophily and influencer networks on song popularity prediction, Scientific Reports (2024). DOI: 10.1038/s41598-024-58969-w


Understanding the synchronization of physiological states during a live music performance
Jul 2024, phys.org

"Music-induced synchronization of heart rate may be the mechanism underlying the coherent behavior of a large audience in a theater." 

Synchronization of physical and cognitive processes is better within an individual compared to that between different individuals; and heart rate synchronization in response to music depends on the reliable physiological responses of the listener, not on their mood or music preferences.

Note that in the study, to guage mood, they had participants in the study listen to the same piece of music on different days. And to quantify the influence of music preference on heart rate synchronization, he investigated whether a person listening to a piece of music selected by the researcher differed in synchronization response to a piece of music that deeply moves them when played in a randomized order.

Why - "From data on small audiences, for example, the degree of proficiency of performers, commercial success can be predicted in terms of reliability."

via Waseda University: Ryota Nomura, Reliability for music-induced heart rate synchronization, Scientific Reports (2024). DOI: 10.1038/s41598-024-62994-0


Classical music lifts our mood by synchronizing our 'extended amygdala'
Aug 2024, phys.org

The study focused on 13 patients with treatment-resistant depression who already had electrodes implanted in their brains for the purpose of deep-brain stimulation. These implants are placed in a circuit connecting two areas in the forebrain—the bed nucleus of the stria terminalis (BNST) and the nucleus accumbens (NAc). Using these implants, the researchers found that music generates its antidepressant effects by synchronizing the neural oscillations between the auditory cortex, which is responsible for processing of sensory information, and the rewards circuit, which is responsible for processing emotional information.

The patients in the study were assigned to two groups: low music appreciation or high music appreciation. Those in the high music appreciation group demonstrated more significant neural synchronization and better antidepressant effects, while those in the low music appreciation group showed poorer results.  

via Center for Functional Neurosurgery at Shanghai Jiao Tong University: uditory entrainment coordinates cortical-BNST-NAc triple time locking to alleviate the depressive disorder, Cell Reports (2024). DOI: 10.1016/j.celrep.2024.114474. 

Intermission on sonic superpowers:
Singing from memory unlocks a surprisingly common musical superpower
Aug 2024, phys.org

They asked people to sing out any earworms they were experiencing and record them on their phones when prompted at random times throughout the day, and found the majority perfectly matched the pitch of the original songs.

"Surprisingly large portion of the population has a type of automatic, hidden 'perfect pitch' ability."

They think it depends on whether the music was recalled deliberately or as a result of an "earworm", which is why they asked people to record themselves as soon as they hear one in their head. So now they think there may be something unique about musical memories and the ways they are encoded and maintained inside our brains.

via University of California Santa Cruz: Matthew G. Evans et al, Absolute pitch in involuntary musical imagery, Attention, Perception, & Psychophysics (2024). DOI: 10.3758/s13414-024-02936-0


Below are a couple articles recalling the genius of Leonard B. Meyer

Bach, Mozart or jazz: Scientists provide a quantitative measure of variability in music pieces
Nov 2024, phys.org

They even mention Meyer in the write up: "According to Meyer, emotions and meaning in music arise from the interplay of expectations and their fulfillment or (temporary) non-fulfillment."

Sorry I have to copy mostly the whole thing, because this stuff is complicated:

They used time series analysis to infer how similar a tone sequence is to previous sequences (the autocorrelation function of musical pitch sequences).

They analyzed more than 450 jazz improvisations and 99 classical compositions, and found the autocorrelation function of pitches initially decreases slowly with the time difference. This expresses a high similarity and possibility to anticipate musical sequences. 

However, they found that there is a time limit, after which this similarity and predictability ends relatively abruptly. For larger time differences, the autocorrelation function and memory are both negligible.

Of particular interest here are the values of the transition times of the pieces where the more predictable behavior changes into a completely unpredictable and uncorrelated behavior. Depending on the composition or improvisation, the scientists found transition times ranging from a few quarter notes to about 100 quarter notes. Jazz improvisations typically had shorter transition times than many classical compositions, and therefore were usually less predictable.

Differences could also be observed between different composers. For example, the researchers found transition times between five and twelve quarter notes in various compositions by Johann Sebastian Bach, while the transition times in various compositions by Mozart ranged from eight to 22 quarter notes. This implies that the anticipation and expectation of the musical progression tends to last longer in Mozart's compositions than in Bach's compositions, which offer more variability and surprises.

via Max Planck Institute for Dynamics and Self-Organization and the University of Göttingen: Corentin Nelias et al, Stochastic properties of musical time series, Nature Communications (2024). DOI: 10.1038/s41467-024-53155-y


Study explores how brain waves reflect melody predictions while listening to music
Nov 2024, phys.org

We aimed to disentangle the frequency-specific neural dynamics linked to melodic prediction uncertainty (modeled as entropy) and prediction error (modeled as surprisal) for temporal (note onset) and content (note pitch) information."

They recruited 20 participants, half of whom were professional pianists, to listen to 10 piano melodies extracted from the work of Johann Sebastian Bach, each lasting approximately 150 seconds, and while being recorded using electroencephalography (EEG).

"An analysis of the temporal response function (TRF) weights revealed that the temporal predictability of a note (entropy of note onset) may be encoded in the delta- (1–4 Hz) and beta-band (12–30 Hz) brain activity prior to the stimulus, suggesting that these frequency bands associate with temporal predictions."

via Max Planck Institute for Human Cognitive and Brain Sciences: Juan-Daniel Galeano-Otálvaro et al, Neural encoding of melodic expectations in music across EEG frequency bands. European Journal of Neuroscience(2024). DOI: 10.1111/ejn.16581

Further Reading for Meyer-like things: