INTERVIEW SERIES: Sonic Thinking: Conversations on Sound, Listening, Culture, and Technology INTERVIEWERS: Gokhan Colak & Arzu Karaduman INTERVIEWEE: Holger Schulze

Three Phases of Sound Studies

You didn’t merely study Sound Studies; you helped institutionalize it through research programs, scholarly networks, publications, and the Sound Studies Lab. Looking back, what has the field become that you didn’t expect it to become?

I started studying at university in the 1990s. At that time, however, there were no study programs focusing exclusively on sound culture and the history of listening. It was not until the early 2000s that we developed the first Master of Arts in Sound Studies at the Berlin University of the Arts together with artists, designers, composers, and sound artists.

This was a time when major monographs by Jonathan Sterne, Karin Bijsterveld, Michael Bull, and Emily Thompson, as well as collected volumes by Veit Erlmann, Michael Bull and others, reached a wider audience. When we argued for this program at the university, however, a common response was that this field might be too niche and too small to have a lasting impact or attract enough students.

Twenty years later, it is clear that sound studies have spread to many universities and cultural institutions. More and more permanent and temporary positions are being announced, and more and more research projects and publications are dealing with the many aspects of sound and listening that are relevant today. Research into sound culture and the emerging practice of sonic critique is becoming increasingly relevant.

From this perspective, the success of sound studies seems both unlikely and unsurprising: unlikely because it is such a minuscule area of expertise, and unsurprising because the widespread use of sound technology, loudspeakers, headphones, and audio formats has been evident since the 1990s. I am confident that sound scholars will continue to research relevant issues and critique sound in the 2050s, 2080s, and likely the 2120s as well. That is, if university culture still exists as it does today and future societies experiencing the hardships of a fully unfolding climate catastrophe still value humanities research.

“Sonic critique can provide guidance for how to understand and how to transform current practices of listening and sounding.”

Sound Studies has developed from a relatively emerging interdisciplinary field into an established area of cultural, media, and sensory research, demonstrating that sound cannot be understood independently from the bodies, technologies, media systems, and cultural contexts through which it is produced and heard. From your perspective, how has Sound Studies changed our understanding of culture, technology, and everyday life?

Since the 2000s, the broader field of sound studies has undergone two significant changes. Initially, sound studies were primarily considered a research domain for historians, sociologists, and musicologists. Many research publications addressed aspects of sound technology, historical accounts, and concepts of listening, as well as their relation to historical and contemporary music cultures. The title of the remarkable inaugural journal article by Karin Bijsterveld and Trevor Pinch documents this focus: “Sound Studies: New Technologies and Music” (2004). Publications by Michael Bull, Jonathan Sterne, Karin Bijsterveld, and Veit Erlmann laid then the groundwork for sound studies, building upon scattered and often idiosyncratic modes of research represented in twentieth-century writings by Jacques Attali, Don Ihde, or Alain Corbin.

Since the 2010s, however, the focus has increasingly shifted toward artistic and anthropological research approaches. Cultural studies of sound have embraced the experiential, sensory, and aesthetic aspects of listening and sounding. Writings by Salomé Voegelin, Marcel Cobussen, Jordan Lacey, Christopher Cox, and my work, “The Sonic Persona” and “The Bloomsbury Handbook of the Anthropology of Sound”, document this phase. Most recently, around 2020, a shift toward an overtly political perspective has emerged, accompanied by a shift toward intersectional and indigenous studies, as evidenced by publications by Dylan Robinson, Nina Sun Eidsheim, Jennifer Stoever, Gavin Steingo and Jim Sykes, Gascia Ouzunian, Pavitra Sundar, and M. Shadee Malaklou.

For me, the actual practices of sound culture and listening research became richer and more complex with each of these three phases. Ideally, this development will continue, leading to a more explicitly formulated mode of sonic critique. My recent research is moving in precisely that direction. In this sense, sonic critique can provide strong guidance for analyzing, understanding and transforming current listening and sounding practices.

“Everything generic will be generated.”

As digital technologies, and increasingly AI, transform the conditions of listening, what new questions do you think Sound Studies now needs to ask?

In recent years, it has become clear that producers and distributors of sound and music production products can profit from commercially accessible LLMs for software automation. Music production LLMs are used in songwriting camps and sound production processes.

As of 2026, it seems that a larger backlash against the surprising overuse of LLMs is growing. Some streaming services exclude LLM products, while others require producers to disclose their use of LLMs. It remains to be seen where music listeners will turn. If they further adopt the use of LLMs, or if they revert to more erratic and complex listening and production practices of, as I would argue, thick listening.

For clearly generic commodities, such as soundtracks for short-form videos, functional music for gym routines, relaxation practices, and a range of shop designs, it might be the most affordable solution to purchase access to a music generation service.

The guideline for automating music production seems to be: Everything generic will be generated. Analyzing these practices of sound productions and how sound cultures transform alongside them will be a major research field in sound studies. This is a clear challenge for sonic critique.

“Thick and complex listening is not pure or idealized; it is sloppy, weird, incomplete, and half-focused. But this ıs how most listeners attend to their environment. It is probably the most realistic and accurate descriptıon of everyday listening today.”

What is Thick Listening?

Pierre Schaeffer’s concept of écoute réduite (reduced listening) proposed a mode of listening in which sound could be approached independently from its presumed source or meaning. Yet contemporary listening is deeply embodied and situated: we listen through bodies, spaces, memories, technologies, and cultural experiences. How would you position your own understanding of listening in relation to Schaeffer’s ideas? Can listening simultaneously be an analytical practice, a bodily experience, and a form of knowledge production?

Schaeffer, Schafer, and various other twentieth-century scholars focused on narrow, refined concepts of expert listening. However, as you rightly point out, listening serves many purposes, often simultaneously. Therefore, I am working with a novel concept that encompasses the greatest possible variety and complexity of listening approaches in one term: thick listening.

Thick listening is neither particularly skillful nor refined. On the contrary, it is erratic, detached, and hybrid. It is sometimes confused and distracted, and other times immersed. It is superficial and sometimes idiosyncratic, and it is a hybrid with multiple layers. It is not just one thing. When performing listening, you combine and hybridize all the modes you and your listening body are capable of. This is everyday life and everyday listening.

You and I embody this thick, complex listening in every moment of our lives. It is not pure or idealized; it is sloppy, weird, incomplete, and half-focused. But this is how most listeners attend to their environment. It is probably the most realistic and accurate description of everyday listening today.

Thick listening puts all the abject, questionable, distracted, impatient, and unfocused modes of listening at its center. These are our most commonly used modes of listening, even if we might not dare admit it. In times of idealized stylizations of listening, amplified by technological models, this realism is crucial and impactful. It is realistic, everyday listening. In this sense, thick listening is a direct example of sonic critique.

“The desire for one-size-fits-all guidelines to sound is understandable, but the reality of listening and sounding is quite different.”

The Plurality Soundscapes in the Digital Age

Raymond Murray Schafer’s concept of the soundscape encouraged us to understand acoustic environments as cultural and ecological formations. Today, however, our soundscapes are increasingly mediated by headphones, smartphones, streaming platforms, spatial audio, notification systems, and algorithmic personalization. From the perspective of your sonic anthropology, how should we understand this contemporary soundscape? Are we moving toward increasingly individualized sonic environments, or are digital technologies creating new forms of collective listening and shared acoustic experience? And what might these transformations mean for our relationship to place, community, and collective memory?

The concept of the soundscape remains artistically and culturally generative, inspiring artists, designers, and listeners in the 2020s. However, the concept was developed in the 1970s. The world, as well as the way we think, listen, and critically reflect on sound, has changed dramatically in the past fifty years. The idea of an ideal, monolithic soundscape relevant across cultures, environments, and historical periods is clearly untenable today.

Contemporary artistic research, on the contrary, focuses on the peculiar character and unique effects of small sonic environments all around the planet, involving a radical and often surprising mixture of analog and digital sound sources, as in the concept of thick listening. In the field of sound studies, the idea of a general, normative set of rules to evaluate sounds is generally regarded as idealistic, unrealistic, and counterproductive.

Certain characteristics, such as noise level, dynamics, complexity, and diversity, can be beneficial in certain situations. However, none of these characteristics are relevant or impactful in all listening situations, cultures, social groups, or historical periods. The desire for one-size-fits-all guidelines is understandable, but the reality of listening and sounding is quite different.

“Meme music ıs the most widely consumed music in the world, transcending genres and cultures, and it is listened to at almost all times of the day and night.”

Meme Music and the New Sonic Vernacular

Michel Chion’s work has demonstrated that sound does not simply accompany images but actively transforms how images are perceived, interpreted, and emotionally experienced. Your work takes this further by treating sonic experience as embodied, material, and culturally situated. In contemporary digital media, immersive environments, virtual production, and AI-generated audiovisual systems, are we moving toward a genuinely multisensory media culture, or are these technologies primarily extending the dominance of visual culture through increasingly sophisticated audiovisual interfaces? And what would it mean to design media from the starting point of listening rather than seeing?

The idealized opposition between listening and seeing has long been dismantled, most notably by Jonathan Sterne, who coined the term “audiovisual litany” in the early stages of sound studies. Take a closer look at your everyday media consumption, conversations, and other activities, and you will realize that new sound technology, sound culture, and listening customs have pervasively entered all areas of your professional and personal life. The harsh distinction is simply a false binary. It is an illusion.

However, a wider awareness of how sounding and listening are indeed intrinsic to many everyday media practices that might seem solely visual at first is still missing. Consider memes, probably the most aggressive and prominent mode of media communication these days.

Memes are most commonly regarded as a visual format. However, with the advent of short-form video and larger bandwidth, more and more memes are time-based. They cannot be separated from gestures, dances, and soundtracks that guide and frame those movements. Consider dances like the “Apple” dance or the “Renegade” dance, snippets of politicians or celebrities misspeaking or stumbling onstage, and the surprising rediscovery of older songs from thirty or fifty years ago selected by social media users for their videos. Musicians present and perform their music online in short-form videos. Political activism and campaigning employ music and bolster their political goals with memes that maliciously mock their political opponents. In my observation, meme music is probably the most important music genre these days. It is undoubtedly the most influential genre of the early 21st century. It transcends genres and cultures and is listened to at almost all times of the day.

These days, contemporary sound cultures across all continents and regional particularities evolve in lockstep with precisely this memecraft. Sounds are employed, distorted, transformed, cut up, and combined in a variety of ways. These are not the methods established in avant-garde composition or sampling in hip-hop or electronic dance music (EDM); rather, these are the methods through which the current sonic vernacular becomes tangible. This new sonic vernacular, emerging before our ears and eyes, promotes accelerated capitalism and serves at the same time as a coping strategy for persevering through its inherent conflicts.

“None of us listens objectively and neutral – not even your favorite LLM.”

Sound, Power, and Algorithmic Culture

Jacques Attali approached music and sound in relation to power, prediction, social organization, and the political economy of culture. If we extend this perspective into the age of platforms and artificial intelligence, algorithms can now predict listening preferences, rank and classify sounds, recommend music, identify voices, and determine visibility, and increasingly generate sonic experiences themselves. Could listening itself therefore become a site of algorithmic power? Who—or what— determines what becomes audible, discoverable, or culturally significant in contemporary digital environments?

Quantified listening is indeed an impactful practice. However, it only applies if you listen to more or less refined sound productions on connected, data-extracting streaming apps. This is one reason why more listeners are returning to vinyl, listening bars, and even music cassettes. Though this is not a mass movement, it is a significant listening strategy for many.

If sonic commodities are an integral part of your business model for maximizing profit, then a stream of automated music and sound productions might be exactly what you’re looking for. It’s about listening to capital and maximizing profits. Many business owners also find joy in these listening practices. This is a major incentive for incorporating LLMs into so many sonic business models. However, LLMs’ listening protocols are neither neutral nor objective, and in this regard, they resemble, quite ironically, the listening of sonic personae like you and me. None of us listens objectively or neutrally—not even your favorite LLM.

“The telos of LLMs is to maximize profit.”

The ROI of LLMs Machine Listening and Artificial Intelligence

In The Sonic Persona you indeed argue that listening is never neutral — it’s shaped by a body, a culture, a history of sensory training that produces a particular kind of listener. Machine-listening systems, by contrast, are often presented as objective: recognizing voices, identifying environmental sounds, transcribing speech, detecting emotions, classifying musical structures, and extracting patterns from enormous sonic datasets. If, as Jonathan Sterne’s work suggests, technologies of sound are embedded within particular historical and cultural systems, what assumptions about human perception are being encoded into contemporary machine-listening systems? What does a machine actually “hear” when it processes sound?

The listening and sounding performed by LLMs definitely does not possess the same characteristics as the listening and sounding performed by a sonic persona. However, there are properties related to the body, culture, and history of sensory experience that influence how LLMs perform listening and sounding. What are these properties?

First, the body concept of an LLM is clearly different from that of more or less humanoid beings like you and me. It is a coded body. There is no metabolism, fear, sexual desire, or obsession with certain activities or sensory experiences. The culture of LLMs is a direct result of the software programming and coding culture from which LLMs evolved. This culture is mainly shaped by young, ambitious, and hyperfocused male students and designers. Significant constituents of a more inclusive culture extending across intersectional areas such as age, gender, abilities, and race are rarely found.

From the outset, the culture of LLMs is pure and radically homosocial; it is hyperhomogeneous. In this regard, LLMs supposedly occupy an objective and neutral position. However, decades of research and cultural critique have shown that such a position is never actually objective or neutral. In most cases, it is a rhetorical function and strategic performance to fortify the powerful position of the speaker.

It is more reasonable to assume that LLMs are artifacts, albeit highly responsive ones, that bear the marks of their initial creators’ and coders’ culture, social behavior, modes of interaction, and mannerisms of expression. Of course, their training data shapes their verbal responses. However, the outer limits of these responses and their constraints are mainly set by their creators and the business environment.

Sure, their training data shapes their verbal responses. However, the outer limits of these responses, or their constraints, are mainly set by their creators and the business environment. In this sense, the goal of achieving a Return On Investment (ROI) is their actual task, for which they were originally developed. This is their telos: They are tools designed to maximize profit. This is their true purpose, their intrinsic mission.

“Sonic pareidolia captivates listeners before they realise it.”

Sonic Pareidolia

Brandon LaBelle’s writings on sonic agency have emphasized the capacity of sound to produce spatial, social, political, and relational effects. This raises an important question in an era of increasingly immersive and interactive technologies: can sound itself be understood as an agent within contemporary media environments? How do sonic experiences shape our movements, emotions, behaviors, and relationships with other people and with space—and what new forms of sonic agency might emerge through AI and immersive media?

Encountering various sonic personae in a given environment is often considered the most common form of listening. Therefore, colleagues here in Denmark have developed the concept of “sonic citizenship” (Marie Koldkjær Højlund and Morten Breinbjerg) to address this issue. You then pose the question: Does sonic citizenship among sonic personae also exist when we join a streaming website, engage in video calls, or produce audio with the support of a cascade of various applications run by LLMs? Many of us interact intensely and emotionally with nonliving entities in our everyday lives – be they smartphones, turntables, bicycles, air fryers, sewing machines, gaming avatars, or the full range of animals, plants, and landscapes that cohabitate with us in our homes, cities, and regions.

Therefore, it is not surprising that pareidolia, i.e. the tendency to recognize faces and characters where there are none, inspires many users to interact with software. LLMs are no different. They don’t even need very sophisticated avatar toolboxes to provide users with enough responses and affordances to firmly enjoy the illusion and wholeheartedly believe that these counterparts possess full agency, character, soul, and even a biography and interest in maintaining a personal relationship.

Sonic pareidolia is more challenging but equally prevalent these days. As soon as a well-crafted voice with enough markers of one’s significant other or of a desired gender directs its articulation towards us, many of us sense deeply, perhaps with a slight feeling of shame, “This voice is speaking to me!” Even if it is a sound production, radio piece, or skillfully produced pop song, a marvelously shimmering gem, many listeners will engage in parasocial relationships, even extreme sonic pareidolia, due to their habits. They believe that this person, sonic creation, artifact, or even the sound of a rushing river, bouncing toy, or speeding vehicle is directed at them. Sonic pareidolia captivates listeners before they realize it.

However, these sounds do not have agency in themselves. Their listeners project agency onto them, and consequently, they have a certain amount of agency. In the strictest sense, sonic critique has the duty to highlight this strange – and, for many, thoroughly joyful – misattribution of behavior related to sonic pareidolia.

Is there Sound in Museums?

You’ve worked as a curator at the Haus der Kulturen der Welt in Berlin, so you’ve had to think concretely about how sound gets staged institutionally, not just theorized. Museums have traditionally privileged visual objects: paintings, photographs, sculptures, and material artifacts. Sound, by contrast, is ephemeral, spatial, temporal, and often dependent on the conditions of listening. With the development of sound archives, immersive installations, spatial audio, and digitally reconstructed soundscapes, do you think museums are experiencing a “sonic turn”? How should cultural institutions rethink the preservation and exhibition of sound when what needs to be preserved may be not simply an object, but an atmosphere, memory, environment, or listening practice?

Museums have indeed been experiencing a sonic turn for some time now. Over the past few years, we have teamed up with colleagues like Alcina Cortez, Eric de Visscher, and Gabriele Rossi Rognoni to map out this ongoing transformation and the critical issues and research gaps associated with using sound in exhibition design.

The most interesting finding from our research is that this transformation is occurring in all types of museums, including natural history, technical history, cultural history, and contemporary art museums. Furthermore, this shift is driven not only by curators or directors who wish to follow current curating trends but also by artists, composers, and designers.

One could argue that in the history of museums, the process of creating an exhibition is moving now even further away from predetermined cultural, scholarly, and pedagogical concepts and toward collaboration with practitioners, listeners, and audience members. The co-creation across these different stakeholders was a striking discovery for me. When our co-edited handbook is available in stores in summer 2027, you will have the opportunity to explore with us the current state of research, sound practices, and experiences in museums in the UK and in China, in Switzerland and the U.S., in South Korea and in Germany, in South-Africa and France, in Italy, Japan, Denmark, and beyond.

“LLMs for the oppressed plebeians and high-end craftsmanship for the affluent leadership class.”

LLMs For the Oppressed Plebeians

Lev Manovich’s work on software, computational culture, and the automation of cultural production provides an important framework for understanding contemporary generative AI. When an AI system can compose music, synthesize voices, imitate performers, reconstruct historical sounds, or generate entirely artificial acoustic environments, how should we rethink authorship and creativity? Where should we locate agency—in the human creator, the model, the dataset, the training process, or the broader cultural archive from which the system learns? And does generative AI represent a new instrument for sonic creativity or a fundamental transformation of cultural production itself?

It is fascinating to observe how the perception of commercial software automation and its results, including all of the famous LLMs, has changed since their ubiquitous advent. The tangible excitement in 2022 has clearly transformed into more complex sentiments among the general public. This division is often quite clear between those who project their hopes of maximizing profits, distinction, and fame onto LLMs and those who experience LLMs exclusively as a massive decrease in the quality of everyday interactions and products from these companies and institutions, which adds to their already overwhelming workloads as employees or freelancers. This division may widen further, as evidenced by the resistance and regulation of the thousands of data centers currently planned for construction around the world. No one actually wants a constantly roaring, humming, wiring, and droning data center next door. The oligarchs directing their construction surely wouldn’t want them built on their tropical vacation islands.

I can foresee a future development in the use of LLMs that closely aligns with what French thinker Régis Debray coined in the 1990s as “The Jogging Effect.” According to Debray, new cultural and technological innovations undergo a radical shift in their broader cultural role over time.They go from being regarded as a promising and exciting future in a technological and distinguished Elysium to being the cheapest option marketed to the general population. This shift makes them affordable and generates massive profits for companies.

The upper and upper-middle classes tend to return to previous technologies that seem to provide a more reliable, valuable, and distinct service. The automotive society popularized jogging and gym memberships. The massive success of streaming services brought back vinyl records, music cassettes, and even tapes. The abundance of inexpensive frozen meals, TV dinners, and canned food made shopping at farmers’ markets attractive. Social media writing revived personal notebooks, sketchbooks, and journaling and kept companies producing fountain pens in business.

I am pretty sure that the classist distribution we can already observe will only become more pronounced. If you live on minimum wage, are unemployed, or work at the lower end of the middle class, you have no choice but to rely on free customer service and low-priced products conceived, produced, marketed, and delivered mainly through the use of various LLMs. However, if you have sufficient capital – perhaps you have inherited substantial wealth and are adept at generating profit from your company’s employees and customers – you will demand and gladly pay the highest prices for personalized service and manufactured or handcrafted goods. In other words: LLMs for the oppressed plebeians and high-end craftsmanship for the affluent leadership class.

“The use of LLMs might only be regarded as nothing more than an activity of low effort, low quality and a marker of low status.”

The Future of Listening: Slop Sounds for Maximizing Profit

Your work invites us to consider listening not merely as the reception of sound but as a way of experiencing, understanding, and inhabiting the world. Looking toward the future, how do you imagine sonic culture evolving at the intersection of artificial intelligence, machine listening, immersive media, digital platforms, sound art, and increasingly automated environments? What forms of listening might become possible that we cannot yet fully imagine—and, ultimately, what does it mean to remain a human listener in a world in which machines are increasingly capable of hearing, analyzing, predicting, and producing sound?

In other words, how will people in the near future experience their world sonically? How will they listen in their everyday lives? At this point in our conversation, I will continue my critical practice of sonic critique by again taking Debray’s jogging effect as a starting point. As Debray argues, any seemingly linear trajectory into the future is subverted and altered by counter movements. In the case of the violent coercion to use LLMs and consume their slop commodities, it’s obvious that modes of production explicitly excluding slop are increasingly valued.

Translating this into the sphere of sound culture and production, one prediction seems more convincing than the rest in light of this sonic critique. Right now, I predict that, in the near or far future, using LLMs for sound production will be seen as an activity of low effort and quality, a marker of low status. In the late 21st century, sonic LLMs might represent stale, odorless, assembly-line products made from the cheapest plastic. Slop sounds for maximizing profit.