Originally commissioned by Festival Novas Frequências in Rio De Janeiro in 2021, DESMONTE is a composition and live performance for modular synthesizer and recorded voice presented by Pedro Oliveira for Between Land and Sea.

A sonic exploration on the afterlives of colonialism and the violence of borders, seeking to uproot voice from body and body from origin. A study on the limits and failures of (machine) listening, the composition appropriates and misuses spectral techniques similar to those of the so-called “dialect recognition software,” a proprietary technology in use for cases of undocumented asylum seekers in Germany. As the voice of death metal singer Fernanda Lira undoes the software’s workings, its timbre undergoes the most extreme mutations, pushed far beyond the comforting perceptual thresholds of recognition. Unfolding as an abstract as much as powerful case for the immanent opacity of the voice itself, DESMONTE articulates a radical critique of the positivist, neocolonial operative logic of biometric technologies.

DESMONTE is a performance for pre-recorded voice and live electronics exploring vocal timbre at the limits of the (juridical) body. The piece is part of a long-term study on the so-called “automated dialect recognition,” a proprietary software in use by the German Office for Migration and Refugees (Bundesamt für Migration und Flüchtlinge – BAMF) since 2017 on cases of undocumented asylum seekers. I’ve been interested in critically inquiring the ways in which this software works, analyzing the gaps of knowledge expressed through what the German authorities say – and what they occlude – in press releases, white papers, and statements available through freedom of information requests. My process enacts a lo-fi “reverse engineering” of each step of this software with the intention of returning its operativities from the mathematical to the listening domain.

One of my hypotheses is that this “dialect recognition” apparatus is not interested in measuring language or dialect, but instead what I call the timbral architecture of the voice. It is by converting snippets of spoken language to a vectorized space, and performing a series of discrete filter operations and feature extraction, that an approximation between the physical characteristics of the speakers’ vocal cords and those found in a larger corpus might reveal an alleged “origin” of the asylum applicant. What is not said is that such measurements occur within the same methodological arrangements of its own design, always already contingent to find that which is designed to measure. The fact that this entire operation is predicated in the interest of detecting “fraud” as quickly as possible reveals discursive details of its own design: to enforce and optimize the asylum process on the BAMF’s side, against the interests of the asylum applicant, with one goal – deportation.

It is through exploring this fabricated connection that the software – and by extension the German border – creates between voice and citizenship, shaped as timbre and origin respectively, that DESMONTE exposes the failures of machine listening to successfully partake, if we follow Sylvia Wynter’s (2003) formulation, in the “overrepresentation of Man”. At the same time, it understands the spaces delimited by such failure to be generative of pathways where it might be possible for the voice to refuse to announce the body, and thus to rethink (or un-think) the figuration of the „human.” I construct the piece entirely with and through the voice of Brazilian death metal singer Fernanda Lira, whose extreme fry scream technique complicates the limits between timbre and voice, noise and signal, identity and identification. In DESMONTE, granular synthesis techniques are used to “zoom in” on her fry scream to derive – rather than extract – signals that take control of decisions done through the audio path, which in turn condition a palette of sounds to emerge, from oscillators tuned to her voice’s partials to realtime spectral filtering and re-synthesis.

My interest in exploring the timbral capabilities present in micro-snippets of articulated fry screams is conceptually and technically geared towards an im/possibility of the body’s disappearance from the voice, while at the same time rehearsing different ways in which the body can make itself present. I see this presentness of the body, articulated through Fernanda’s fry scream, as being disconnected from extractive practices that allow for measurability. Rather, exactly by exceeding the body and making it overtly present – both as throat pulsation and as spectral figure – that a disconnection from body and the instrument of identification is exacerbated.

Death and Black metal’s long history of scream techniques and skill notwithstanding, DESMONTE emphasizes the ways in which the speaker/singer is represented in such genres; I am speaking here namely of the figure of an “extreme otherness,” predicated on Death and Black metal’s imagetic (and sonic) relationship to monstrosity, which immediately positions its fry scream away from more traditionally “academic” extended vocal techniques usually found in electroacoustic music and sound art. However, my choice of working specifically with Fernanda’s voice also emphasizes the politics that condition this figure within a mostly masculine (and oftentimes misogynistic) musical genre to extend it to more encompassing conversations towards a feminist and anticolonial “monster.”

This mode of reassembling the “monster” – not visually but sonically/haptically – through the extremely skillful physicality expressed by her voice, refuses to announce anything but the body “in the raw,” (Ferreira da Silva 2018) that conditions the possibility of that physicality, always against identification, measurability, and taxonomization. With that, Fernanda’s extreme fry screaming and the spectral narrative constructed in DESMONTE hold the possibility of exceeding the body by insisting on – and in some ways enacting  – the separation between body and origin, and origin and citizenship that systems such as the BAMF’s dialect recognition software seek so boldly to sediment as one and the same, biologically immutable, always available for evaluation.


Ferreira da Silva, D., 2018. In the Raw. e-flux.

Wynter, S., 2003. Unsettling the Coloniality of Being/Power/Truth/Freedom: Towards the Human, After Man, Its Overrepresentation–An Argument. CR: The New Centennial Review 3, 257–337.

Produced with the support from the Akademie der Künste Berlin, with funding from the Federal Government Commissioner for Culture and the Media as part of NEUSTART Kultur.

Voice: Fernanda Lira

Dramaturgy: Giuliana Corsi

Curation: Mattia Capelletti

Presented in collaboration with Cripta747 as part of The Listeners. Realized with the contribution of Ministero della Cultura, Goethe Institute and Kultur Ensemble.

With the support of: