December 12, 2021
It’s a cold January night in Jangpura, Delhi. I am sitting in the living room, broadcasting the sounds of my neighborhood to the world. Broadcasting is not the technically correct term in this context—I am actually streaming. And I am not really streaming to the world, only to a couple of friends helping me test the installation of an Icecast[1] server. Music playing on loudspeakers from the settlement of Tamil migrant workers across the railway tracks cuts through the thick Delhi smog into my microphone and out through the public yet unlisted link. It’s oldies night tonight, and we are listening to an Ilayaraja number—“Nila Kayuthu” from the 1982 film Sakalakala Vallavan. I feel like I am eavesdropping here—I am an incidental listener, far from the source of the sound, not just in terms of physical distance but also in terms of a psychic distance from a culture that I can only access as an echo. And an echo of the echo streams over the internet as a sound-image.
Internet radio is a bit of a misnomer. Radio in the traditional sense of the word brings to mind a realm of modulating signals, airwaves, and antennae. Streaming instead is a game of continuous digital data transfer using encoding techniques that discretize data into packets in anticipation of a simultaneous decoding at the other end of a network. Technically streaming needs to be contrasted with downloading, where an entire media file needs to be obtained by the user before it can be played. Instead you plug into a livestream to listen, and it perishes when you unplug. Setting up internet radio then is a bit like laying down a pipeline. You start by connecting pathways for sound to first reach a computer, then from the computer to the local network and from there to the public internet. This last hurdle is especially tricky, because it requires slipping through various protocols and bureaucratic forcefields that govern and regulate the flow of data over the internet. If everything works out correctly, then you get to hear the sounds playing on your computer on a link that you can share with friends. The only evidence of a listener is more data in the form of server statistics. My server is an old, noisy laptop, perched on the table next to me, in constant danger of a cat jumping on it. It hasn’t been shut down in the past few days, and its innards have been churning with multiple installation and debugging sessions. The computer is tired. Its cooling fan buzzes louder whenever someone visits the stream—that’s the other evidence of a listener.
English—unlike Marathi, the language that I grew up with—has two words to describe the everyday act of auditory reception: hearing and listening. Le Mardi Gras Listening Collective tries to distinguish between the commonplace meanings of the two terms—hearing as a passive physiological phenomenon and listening as an active skill, a practice to be learned. For example, “to really listen” is understood as a good trait, and “listen to me” does not have the same shade as “hear ye.” The former is a directive toward obedience, while the latter is a proclamation on the senses, an imposition that denies you a choice in the matter—a speech act. The “can you hear me?” from a patchy video call is more a confirmation of signal reception than an inquiry about the interpretative capacity of the person on the other end. In this schema, the notion of “work” is located squarely in the realm of listening. We seem to have to learn how to listen and, in that process, become entangled in various forms of systematization—ways of distinguishing between good and bad sound, symbolizing them, adjudicating notions of correctness and fidelity, inventing control mechanisms and forms of disciplining the body toward becoming a listening subject. A “good” listener filters out the raw, unnecessary information absorbed by the senses and makes meaning out of it. A good listener of an internet radio station such as this improvised, DIY assemblage would then have to test their skill against the lags, silences, interruptions, and disruptions built into this apparatus by the very nature of interdependencies leading all the way down to the threat of the cat jumping on the computer.
In contrast the collective poses the question: What does the sound of music do to the sense of hearing? This bypasses listening altogether and focuses squarely on hearing as capacity. It posits a sense of hearing that is not passive or inert, but malleable and reactive—reactive to the world in a way that is shaped by the conditions of life around it. Furthermore it engages musical sound as a mode of contributing to this continuous process of the territorialization and deterritorialization of the senses. The work of hearing then could be imagined as technical—in the sense that it mediates the liminal space between a psychic interiority and a material exteriority and that it can be affected, recomposed, tuned. To borrow a term from Lacan—extimacy, “something that is strange to me, although it is at the heart of me”[2]—would also imply a techno-diversity of hearing. Imagine different kinds of ears—stretchy ears, ears that can catch sounds from vast distances, ears that can trap echoes, ears made up of other ears, ears tuned differently.
In a work titled Ruptures, Kerala-based artist V.V. Vinu makes thousands of ears with terra-cotta, which cling to each other in the form of beehives.[3] These beehives hang from carved wooden beams of the othalam tree—also known as the suicide tree, because its poisonous fruit has historically been used for this purpose, especially by the marginalized caste communities of the region. The carvings on the beam form a chain of Malayalam letters, floating above the ground like speech bubbles in a graphic novel. The text itself transcribes the onomatopoeic sound of loud belligerent laughter—“Ha ha ha ha ha ha haaaaa aaaah.” Vinu recalls childhood nights spent watching local theater performances that narrate mythologies of the so-called righteous daevas’ conquests over the rakshasas in battle—moral stories about the victory of good over evil that work to erase historical caste injustices. But even though they are slayed, the rakshasas’ defiant laughter doesn’t die. It still reverberates into the night and in Vinu’s ear. It becomes a force in itself, a sound effect that carries with it many, many ears as it moves. Instead of hearing as a way to sense sound, sound is a means to alter the logic of sensibility itself and, by extension, to think of sensibility itself as a result of cultural production that is always a condition shaped and enabled by social life.
There is a fascinating image that Franz Fanon renders in This Is the Voice of Algeria[4]. He describes a scene from a household in an Algerian village during the struggle for freedom from the French colonial occupation in the 1950s—members of the household are listening together to a clandestine radio broadcast by the revolutionaries hidden in the mountains. There is hushed silence in a crowded room while those closest to the radio receiver try to tune in to the rumored frequency of the broadcast. The radio itself speaks in alien tongues, spitting noises and static drones, hovering in and out of legibility. Bated breath till the broadcast ends, then an intense discussion that goes on through the night, arguing over interpretations of what was heard, the latest news of the revolution. Disagreements, confirmations, confrontations. Unsettlement of meaning.
Somehow in Bhogal Jangpura, a small neighborhood in South East Delhi historically made up of migrants and refugees, the soundscapes can be quite heterogeneous—wedding brass band practice sessions early in the morning, Afghani and Uzbek satellite TV channels playing pop music in the market. My favorite momo (dumpling) stall is run by two young rappers from Nepal. Sometimes, legato rhythms of the parai drum resonate in the air to signal a death across the railway tracks. Two white horses gallop along the street every morning at nine. Nights carry the sounds of vehicle horns and screeches and the strange groans of the railway line and neighbors cooking and fighting and taking late-night walks and snoring. The sound of the music coming from outside my window has now become familiar, although it’s in a language I don’t understand. The question then for a potential internet radio infrastructure, for a technical apparatus that can mediate sound live across time and space, and for a practice of making and carrying sound, is a question of delinking the work of hearing and listening. How can such an assemblage intervene in our listening habits and recover a social feeling out of sound? How can it help us discover new modes of hearing? How can it affect the bendiness of ears?
After a while my friends drop off, and it’s just me, listening simultaneously to the radio and to the soundscape—earphones in one ear. A split listening.
May 22, 2022
“The image is always incomplete. It is always porous. Even though it may be framed in our own ideological frameworks, it always bleeds,” says Santhosh Sadanandan, in the middle of an intense discussion after a lecture performance by Mohit Shelare at our shared studio in Kashmere Park. I remember sitting in this room, broadcasting this conversation, in a circle around stacks of photocopied drawings depicting images of the targeted pogrom of Muslim communities in North East Delhi from February 2020—like mourners around a tomb. The images were sourced from various news sources and social media reportage, digitally copied in monochrome using drawing tools on an iPad, then printed and repeatedly photocopied on A4 sheets of paper. What used to be photographs are now photocopied sketches. Mohit talks about how the past few weeks have been drenched in these images through social media and TV, and he has a desire to draw a texture out of them. You can see the artist’s hand in the way that the familiar images are rendered. You can also see the ink stains and ghostly lines of the photocopier. Often multiple frames occupy a single sheet, like the layout for a storyboard or a graphic novel, but without any text. The debate is at a critical point, and the difference between an “image of violence” and a “violent image” is being dissected. Santhosh alludes to an older philosophical problem of the status of the image and its relationship to reality—as substance or as likeness. Are images mere copies of reality or do they hold a part of that reality in themselves? Are they imitations of the world or do they actively shape and create it? And if they do, then is the reality that they participate in different from ours? The classic example of Orson Welles’s 1938 broadcast of H. G. Wells’s War of the Worlds illustrates this tension. Panicked listeners thought that the Martians were really here to take over the world. How should we read this incident? Is it a trivial instance of misinterpretation of context, or perhaps did the image, simultaneously mediated by the radio into many receivers, manage to touch a nerve—that just for a few hours, a small part of the world experienced an expression of a deeply internalized paranoia toward an alien other that always already existed under the surface? This possibility—that images form a reality of their own, not as degraded copies pointing to stable notions of indexicality, but as part of a world of images in motion, eager to attach themselves to new meanings—that the true value of the image is not to be found in what it claims to represent, but in the way that it operates in this world, and that this betrayal is also its secret power. The work of an artist is then not just intervening in different modes of production of images but also enabling different modes of exchange, forums, and arenas of interpretation around them. To move is to be, and to move thought is to make it. It’s the first time we are broadcasting remotely, away from the server. The decision is quite impromptu. A poster with a link is circulated across a bunch of WhatsApp groups. Friends text back, trying to identify people in the room. Every so often the discussion is interrupted by an inquiry about the listeners on the stream. Eventually the room takes over the broadcast. Intensity disperses into a texture of overlapping voices. We serve everyone roohafza[5] and gin. Conversations go on long after the broadcast ends.
While on a trip to Kashmir later that year, we ask a person running a mobile repair shop in Charar-e-Sharif how he listens to music with all the internet shutdowns.[6] He unveils a whole infrastructure of torrents and flash drive file transfers that enables local music production and circulation in the valley. He is the provider of music for the whole village. He fills up people’s phones with gigabytes of music downloaded on his laptop, updated every couple of weeks. We fill ours as well. Uzma Falak writes about how the ear is shaped in Kashmir[7]—how it learns to discern the sounds of an AK-47, a light machine gun, and a self-loading rifle; how it differentiates between army and militants based on firing patterns; how it endures the tortured screams that the army amplifies over loudspeakers all through the night to terrorize the village. She describes an ear that can distinguish between different kinds of silences—an ear born out of contingency. Our friend and artist Khursheed Ahmed tells the story of how you can tune in to both Radio Pakistan and All India Radio and how both stations announce seven o’clock news every day—half an hour apart. His playlist of Kashmiri music plays every day on radio roohafza.
With -out-of-line-[8] we facilitate a workshop for Ambedkar University students. It’s the second wave of lockdown. Physical classes are suspended at the university. It’s already been nearly two years since the pandemic began, and there is no end in sight. We are starting to get used to waiting by now—for vaccines, oxygen cylinders, and remdesivir, for graphs to plateau; waiting in hospital waiting rooms and Zoom waiting rooms. We meet online twice a week, and each time we bring recordings for the class to listen to. We notice what Zoom does to sound—how it struggles to filter out the pressure cooker whistling in the kitchen, the foghorn of the passing train, or overlapping voices. We notice how our voices sound different over Zoom, and different yet again over Google Meet and WhatsApp. The pandemic forces the classroom into our domestic spaces. We suddenly get to know each other’s cats and roommates. We talk a lot about the experience of time in tense isolation, about the things we have heard around us and the things we miss hearing. We talk about how soundscapes are controlled and domesticized, and how a sound recording can become an expressive object. One night I read This Is the Voice of Algeria on the radio livestream as a bedtime-reading broadcast. Students respond with tracks of their own.[9] They record their cats and their keyboards and their clocks and their conversations; they sample TV news broadcasts, YouTube videos, and Instagram Reels. They mix and layer them into sonic montages. We host a listening party on radio roohafza and broadcast all the tracks. Then we all go live. Everyone talks about where they’re tuning in from—different cities and spaces; some listening with friends and family, others alone; on earphones or Bluetooth speakers; on a walk, while cooking or eating or napping. I ask how it felt to listen to their own work through this medium. It was different, they say, because we know that others are listening along with us.
December 25, 2021
The Indian prime minister has his own radio show. His voice is broadcast every week, constructing a sonic image of the leader addressing the nation. At the same time, on YouTube, you will find dozens of imitation artists teaching you how to sound like Modi.[10] They all have extremely specialized teaching techniques. The tone, texture, and diction of the voice is carefully deconstructed to the level of phonetics and physiology. One tutor teaches you to vocalize the sound of a cow before graduating to Modi’s voice. Another one asks you to breathe as if you’ve just come from a long run. This leads to some hilarious situational comedy—almost all the Modis ask you to like, share, and subscribe to their respective channels. They are all simulacra, in the sense that there is no original. The original is absent—because the leader’s voice itself is a mediated construct. AI speech models work on a similar premise. They analyze vast amounts of audio recordings to construct a latent space of abstractions around the voice, made up of multiple dimensions, each coinciding with a vocal attribute—pitch, tone, cadence, phonetic structure—to generate a new yet uncannily familiar voice, an endless series of resynthesized leaders. You can make them say anything you want.
On April 14 Mohit, Aasma and I dig through YouTube to search for recordings of B. R. Ambedkar’s voice and do an Ambedkar Jayanti Special broadcast on radio roohafza.[11] We can barely find forty-two minutes of material: a speech delivered to the Constituent Assembly, an interview with the BBC, and a noisy recording from Kathmandu of a speech given two weeks before his death, titled “Buddhism or Communism.” Along with the poster, we circulate a note:
This broadcast consists of a collection of recordings of Ambedkar’s voice found on YouTube, arranged in a sequence. The voice in these recordings comes from different times, locations and situations in Ambedkar’s life. Each time it is heard, it constructs a new imagination of a listener that exceeds those that came before it. In that sense, this voice still resonates in today’s social and political reality and continuously revolutionises the idea of collective living and equality.
There must be, and we think there are voices from Ambedkar and Ambedkarites that have not been recorded, or even heard. To whom we listen is an important choice—rather a decision that is facilitated by flows of power within mediatic environments. This presence of Ambedkar’s voice also provides that which is not voiced yet, and provokes how to listen to the absence of a voice.[12]
August 31, 2022
We are reading Etel Adnan’s Arab Apocalypse live on the radio.[13] It’s a long poem with fifty-four chapters, and it elicits a two-and-a-half-hour-long round-robin between Yamu and Jad in Beirut and Aasma and me in Delhi, which starts feeling a bit hypnotic once you’re deep into it. Before starting Yamu apologizes for the inability of our voices to articulate the drawings, marks, and symbols that become part of Adnan’s text—unpronounceable signs on the page that can only be encountered visually. Instead the voices bring something that the texts can’t: Beirut traffic mixed with the now-famous Jangpura train foghorns, our stutters mixed with the stutters of the supply chain of sound through the server. She does not apologize for the infrastructural disruptions though, but instead asks us to consider them as opportunities to think of the real conditions of cultural production itself.
Yamu addresses “All listeners there around with us.” We don’t know who is listening, but we know that someone is. If traditional radio is a mass medium, internet radio is a database medium. Nonetheless they all construct listeners in their imagination. They address their unseen audience as if they were a singular entity. The voice on the radio, disembodied and pervasive, creates an imagined intimacy. The listener on the other end is never alone, but a part of a whole. The listener of radio roohafza is incidental, both in the sense of an untimeliness and in the sense of passing through. More often than not they stumble upon the link without knowing where it will lead. In that sense radio roohafza is not a platform, but an extrusion. It appears to form temporary assemblages with networks, bodies, and voices, and then disappears to reappear elsewhere, another time. What links this nomadic movement is a disposition toward making new forms of association. The listener has to change with every broadcast, then be reimagined along with new protagonists. The question of delinking hearing and listening has to be continually readdressed—a recursive spiral that brings in new imaginaries with each loop.
July 20, 2023
I am in Studio Lang de Moun Mon, a recording studio inside a Haitian community space in the Observatory neighborhood of Johannesburg, surrounded by people I’m meeting for the first time: Soweto jazz guitarists, art school students, pedalgazers, bedroom Amapiano producers, sangomas, poets, rappers, kids doing vlogs for YouTube. It’s a lively crew. I’ve been in residency here for the past few days, hosted by Mzu Nhalabati and Wandile Ndlovu, musicians and audio engineers at the studio. Over the past week they have been schooling me in South African jazz histories, the intricate differences between gqom and kwaito and how to identify the Durban sound. They’re my guides to the city, to its gigs and jam sessions and guitar shops and record stores. In turn I tell them about radio roohafza—you know, it would be cool to do a live broadcast from here.
We go live on a cold July afternoon, mellow embers crackling in the fireplace. We play for hours, stream for hours. The date is carefully selected after going through the load-shedding schedules of the electricity provider. It’s Felix’s birthday today—a cake appears. Happy birthday on the radio. Kerline serves lasagna, then belts out a banger. Before we start someone asks me what this roohafza thing is all about.
It’s an internet radio station. But not really; I mean it’s just a laptop running in a shared studio in Delhi, to be honest. It’s like this experimental setup, really small scale, very low key. It will mostly be friends listening, and friends of friends—in Delhi, Cape Town, Beirut, Berlin, Hong Kong, Dubai, and Bengaluru—those are the ones I know for sure.
By the end of the night, I feel like saying that it’s just an excuse.
It’s August 2025 at the time of writing this epilogue. I started working on this text in December 2023, first in the form of a presentation for the conference Modulating Realities: Networks of Sonic Thinking (December 14–15, 2023) organized by Liquid Architecture and Sarai at the Centre for the Study of Developing Societies in Delhi. Since then this text has been evolving along with a continuous stream of feedback from the editors and reviewers at the Stedelijk.
The thoughts rendered in this text draw from a theoretical scaffolding of many thinkers through the fields of social theory, media infrastructures, sound studies, and the philosophy of technology. From the work of Jacques Rancière, I draw the concept of the distribution of the sensible—the idea that common social worlds are built on partitioning and arranging the perceptible (what can be seen, heard, said, and done) and that politics arises when this order is interrupted, reconfigured, or redistributed. This idea finds resonance in the distinction between listening and hearing made by Le Mardi Gras Listening Collective, who argue for hearing not as a passive sensory process but as a capacity that is actively shaped and reshaped by the social realities surrounding it. It also finds purchase in my reading of Franz Fanon, especially in the context of his thinking around the psychosocial dimensions of technology. More recent work by Fred Moten and Stefano Harney on ideas around fugitivity both as a refusal to demands of legibility and as an improvised state of being with others has been central to the decisions made while building and operating these servers and to reconcile the aesthetic and political dimensions of this work.
I borrow the idea of interplay from Keller Easterling, who uses the word to point to the relational field between things, rather than the things themselves. Interplay occurs in the active spaces of exchange, adjustment, and repetitions between and within object forms. I also draw from Manuel DeLanda’s reading of Deleuze and Guattari’s writings on the assemblage, and AbdouMaliq Simone’s ideas of improvisation and people as infrastructure in the context of reconciling social activities and technical activities—a theme that has constantly been on my mind while working on these servers and this text. To think the computer, the internet connection, the operator, the voice, the listener, the cat, the friend that forward links to others—that each of these are parts of temporary and improvised assemblages, to think of broadcasts as relationships. This dispositional thinking that collapses the boundaries between technical practice and social practice has been at the core of this text.
Bernard Steigler’s ideas around technics as exteriorization of memory—as materially bound tertiary retention—complicate both the question concerning technology and the question concerning psyche. For me this formulation resonates with Lacan’s concept of extimacy—the most intimate also being the most external. The kernel of the subject is located outside, in the other, in language, in relations, in exteriorized memory. Throughout the process of writing this text, I have been paying close attention to how the server and its activities are spoken about, referred to, in formal forums and informal conversations, in broadcasts and otherwise. These might read as anecdotal incidents on one level, but for me they are precisely sites of subjectivization, however minor and inconsequential. To recognize them as such and give them form has been one of my aims in this text.
This text has been developed while constantly in conversation with colleagues and cothinkers in Delhi, especially Shveta Sarda, Santhosh Sadanandan, Aasma Tulika, Suvani Suri, and Jeebesh Bagchi. A formulation of the idea of “chatter” has been brewing through these discussions—as a nondominant discursive flow between selves, milieus, and things that precedes modes of legitimization and authorization. One of the key questions that emerges in relation to this discussion is the relationship between chatter and infrastructure. A shared self-hosted server becomes an interesting technical object to think through them and work them out, hopefully in the near future.
Finally I would like to say something about the form of this text. During the many rounds of editing and peer review at the Stedelijk, the question of “evidencing my engagement with the literature of the discourse” has reappeared again and again. For the most part the thinkers and their ideas mentioned above are not directly cited in the main body of the text itself, but rather live in a subterranean register of the text—not fully on the surface, but visible through cracks, faults, minor exposures, and signaling tactics. This is deliberate, only because I wanted to accentuate and amplify the affective and sensorial dimension of the ideas presented in this text over the legible and legitimizing function that citations often perform. A constant thought haunting me while writing this has been how to render ideas and experiences as images—thought-images that operate, after Guattari, not just by the logic of intelligibility but by the logic of intensity. It has been an extremely rewarding experience trying to work out the rhythms, flows, cuts, and dynamics of the text from this vantage point. This is where the anecdotal and the incidental are put to work. This is not a question of obfuscation of source—there is nothing to hide—but rather of internalization of ideas and exteriorization of a voice in singular ways so that I can (to borrow from Shveta Sarda) be held accountable for the mess they make on the page.
Kaushal Sapre is an artist based in Delhi. His work has to do with rethinking the self and the social in everyday technics. He develops web servers, manages an internet radio station, writes software, teaches courses around art and technology, stages images and objects, makes and performs with experimental sound instruments. He has exhibited and contributed to events at Fondazione Sandretto Re Rebaudengo, Madrid (2024); Watermans Art Center, London (2023); Kunsthalle Bern (2023); Hangzhou Triennale for Fiber Art (2022); Rockbund Art Museum,Shanghai (2022), Five Million Incidents, Delhi (2020)
[1] Icecast is a free/libre and open-source software designed for the purpose of streaming media over the internet.
[2] Jacques Lacan, The Seminar, book 7, The Ethics of Psychoanalysis, 1959–60, trans. Dennis Porter (London: Routledge, 1992), 71.
[3] See V.V. Vinu’s Ruptures on the Shrine Empire website (accessed December 10, 2025).
[4] Fanon, Frantz. “This Is the Voice of Algeria.” In A Dying Colonialism. New York: Grove Press, 1965.
[5] radio roohafza takes its name from roohafza—a drink popular across the Indian subcontinent, first made in Delhi in 1906 by Hakim Abdul Hameed. Rooh-Afza literally translates to “soul refresher.” The specific Unani recipe of roohafza combines various cooling agents and is believed to be an excellent remedy against the hot, dry winds that blow across the subcontinent in the summer.
[6] Jammu and Kashmir has had the dubious record of having the largest number of internet shutdowns in India. The now–union territory has witnessed three out of the four longest internet shutdowns and communications blockades in India.
[7] Uzma Falak, “Echoes from an Anechoic Chamber,” in “Capture All: A Sonic Investigation,” ed. Laura Mclean and Mehak Sawnhey, special issue, Disclaimer.
[8] -out-of-line- is a collaborative practice invested in creatively participating in the socio-technical experience of communication. Since August 2019 -out-of-line- has been organizing interventions in and around telecom infrastructures with an evolving cohort of artists, designers, researchers, programmers, and musicians. Recent work has involved playing with the form of invitations to think together about connectedness and publics.
[9] radio roohafza, “radio roohafza broadcast #4—Listening Party,” YouTube, 2 hours, 35 min., May 21, 2022.
[10] “radio roohafza broadcast #1—M for Mimicry,” This Might Not Be, December 25, 2021.
[11] April 14, the birthday of B. R. Ambedkar—activist, politician, thinker, and leader of the oppressed castes—is celebrated across India as “Ambedkar Jayanti.”
[12] radio roohafza, “radio roohafza broadcast #5—Ambedkar Jayanti Special,” YouTube, 42 min., 45 sec., May 21, 2022,
[13] Etel Adnan, Arab Acocalypse, trans. Jalal Toufic, 3rd ed. (Sauasalito, CA: Post-Apollo Press).
