Vincenzo Madaghiele

• Madaghiele, Vincenzo; Fasciani, Stefano; Kelkar, Tejaswinee & Erdem, Cagri (2026). Modeling Relations Between Musical Events in Continuous Time with Transformer Models for Live Co-Improvisational Interactions | Zenodo. I Coelho, Guilherme (Red.), Proceedings of International Conference on AI and Musical Creativity. AI Music Creativity (AIMC). doi: 10.5281/zenodo.22276042. Fulltekst i vitenarkiv Vis sammendrag

  This paper presents a live semi-autonomous system that co-improvises with a live musician using a model learned from the relationship observed in paired recordings of an improvising duo. This responsive system is based on a deep learning approach that models the relations between sequences of events produced by co-playing musicians in continuous time, using transformer models. We present the architecture for simultaneous sequences of sonic events and provide a quantitative evaluation on several representative tasks. Results are compared against a canonical transformer and a multichannel Factor Oracle, the latter being a widely used model for live sequence-based symbolic music generation. We detail a live implementation employing concatenative synthesis and introduce a customization procedure in which the generative model is iteratively retrained on curated, satisfactory sections from sound recordings of actual human-system co-improvisational interactions. This iterative fine-tuning enables the model’s stylistic output to diverge from the original training corpus. Two musical use cases demonstrate the application of this technique.

• Madaghiele, Vincenzo; Fasciani, Stefano & Erdem, Cagri (2026). Personalising behaviours of the multi-agent autonomous looper with a corpus of annotated co-performances. Organised Sound. ISSN 1355-7718. s. 1–12. doi: 10.1017/s1355771826101484. Fulltekst i vitenarkiv Vis sammendrag

  This article introduces a method to personalise the behaviour of a music co-improvisation system using a genetic algorithm that iteratively identifies optimal configurations governing the system’s behaviour according to a musician’s aesthetic objectives, using a corpus of annotated co-performances. The corpus annotations represent the decisions the semi-autonomous system should learn to make in different musical contexts, reflecting the musicians’ styles, ideas and aesthetic preferences. We apply this approach to personalise the multi-agent autonomous looper (MAAL), a co-improvisation system that autonomously samples and loops segments of an improvised performance by integrating machine listening with a rule-based framework. After detailing the design of the evolutionary search algorithm, we present the results of a study involving five expert improvising musicians. These musicians applied the proposed method to personalise the MAAL and co-improvised with it. We assessed the extent to which the MAAL’s behaviour, adapted to their personalised data, aligned with their individual aesthetic preferences. Findings show that automating the functions of sampling and layering loops seems to invite explorative and responsive creativity and that personalisation can facilitate co-adaptation between musicians and artificial agents by establishing a defined aesthetic vocabulary of co-creative behaviours.

• Madaghiele, Vincenzo; Fasciani, Stefano; Kelkar, Tejaswinee & Erdem, Cagri (2026). Autonomous Listening-Based Synthesizer Control with Reinforcement Learning for Live Sound Matching. Proceedings of the International Conference on New Interfaces for Musical Expression. ISSN 2220-4792. s. 751–758. doi: 10.5281/zenodo.20784273. Fulltekst i vitenarkiv Vis sammendrag

  In this paper, we build on a previously proposed method for training an artificial agent that performs live control of sound synthesizers, using reinforcement learning and machine-listening to match the sound of a live musician. The agent listens to an incoming live audio stream and attempts to match its sound according to specific audio descriptors by controlling the parameters of an arbitrary synthesizer. We introduce a reinforcement‑learning model with a continuous action space that integrates sound synthesis during training and live performance. We propose five complementary reward functions for matching target sounds, and evaluate their effectiveness in three scenarios: in‑domain matching of synthesized target sounds, out‑of‑domain matching of synthesized targets, and out‑of‑domain matching using a corpus of musician‑recorded audio. Finally, we compare the proposed reward functions across these scenarios, we discuss a live implementation and we present the results of quantitative experiments evaluating the matching task.

• Madaghiele, Vincenzo; Lund, Leonard; Holzer, Derek; Kelkar, Tejaswinee; Tatar, Kıvanç & Holzapfel, Andre (2026). Expanding the machine: Notating generative synthesis with a state-based representation and a navigable timbre space. Organised Sound. ISSN 1355-7718. doi: 10.1017/S1355771825100915. Fulltekst i vitenarkiv Vis sammendrag

  Notating electroacoustic music can be challenging due to the uniqueness of the instruments employed. Electronic instruments can include generative components that can manipulate sound at different time levels, in which parameter variations can correlate non-linearly to changes in the instrument’s timbre. The way compositions for electronic instruments are notated depends on their interfaces and the parameter controls available to performers, which determine the state of their sound-generating system. In this article, we propose a notation system for generative synthesis based on a projection from its parameter space to a timbre space, allowing to organise synthesiser states based on their timbral characteristics. To investigate this approach, we introduce the Meta-Benjolin, a state-based notation system for chaotic sound synthesis employing a three-dimensional, navigable timbre space and a composition timeline. The Meta-Benjolin was developed as a control structure for the Benjolin, a chaotic synthesiser. Framing chaotic synthesis as a specific instance of generative synthesis, we discuss the advantages and drawbacks of the state- and timbre-based representation we designed based on the thematic analysis of an interview study with 19 musicians, who composed a piece using the Meta-Benjolin notational interface.

• Demir, Arife Dila & Madaghiele, Vincenzo (2025). Reinterpreting Anatolian rugs through multi-sensory experiences. Biennial International Conference for the Craft Sciences (BICCS). Fulltekst i vitenarkiv Vis sammendrag

  Anatolian rugs convey layered stories and meanings through their colourful motifs, reflecting both cultural traditions and the personal interpretations of the weavers. In our research, we explore alternative ways to communicate these narratives through multi-sensorial experiences, creating immersive engagements. Using interactive connected smart (ICS) materials such as sensors and microcontrollers, we develop a multisensory rug that mediates touch and sound interactions. In addition, we emphasize the importance of acknowledging our situatedness as makers: whether we come from within or outside the culture in which we work, our backgrounds and tools shape how we approach craft traditions. Thus, we reflect on how we can remain respectful to the traditions while reinterpreting them through our positionalities and contemporary methods.

• Madaghiele, Vincenzo; Fasciani, Stefano; Kelkar, Tejaswinee & Erdem, Cagri (2025). MAAL: a multi-agent autonomous live looper for improvised co-creation of musical structures. I Bown, Oliver (Red.), Proceedings of International Conference on AI and Musical Creativity. AI Music Creativity (AIMC). doi: 10.5281/ZENODO.16946080. Fulltekst i vitenarkiv Vis sammendrag

  Live looping is a musical practice based on recording, repetition and layering segments of a live music performance. Layering loops is a technique used to create musical structures that evolve over time throughout a performance, and it is particularly popular among solo improvising musicians. This paper presents the Multi-Agent Autonomous Looper -- a system that autonomously selects musical segments to be sampled and looped during a live performance, using a set of rule-based agents. Each loop track is managed by an agent that listens to the musician's performance and to the concurrently looping segments on other tracks. The decision on which segments of the performance to loop depends on user-selected rules based on comparison of musical properties, expressing varying degrees of sonic compatibility. This approach characterizes the entire system as a multi-agent network capable of autonomously generating emergent musical structures throughout a performance. The Multi-Agent Autonomous Looper aims to expand the possibilities for indirect control and co-creative interactions in live looping for improvising musicians. In the paper, we detail the system's objectives and design, its offline and online implementations, and discuss specific settings and behaviors across three representative use cases.

• Madaghiele, Vincenzo & Demir, Arife Dila (2024). Pain Creature: interdisciplinary collaboration in the design of an embodied textile instrument for interactive dance. Proceedings of the International Conference on New Interfaces for Musical Expression. ISSN 2220-4792. 68, s. 465–473. doi: 10.5281/zenodo.13904917. Fulltekst i vitenarkiv Vis sammendrag

  This paper describes the development of an interactive textile instrument called Pain Creature. The instrument is the result of a collaboration between a sonic interaction designer and a textile designer. Through textile and auditory qualities, Pain Creature explicates different aspects of the second author's experience of chronic pain. The instrument can be used as a tool to reflectively engage with the user's pain experiences and as a performance instrument. This paper introduces the collaborative design process employed to develop the artifact, focusing on how contributions from sound design, textile design and movement practice were combined in the design phase. The paper describes the technical design of the artifact and discusses how knowledge from different disciplines affected the development of the instrument and its use in performance.

• Madaghiele, Vincenzo & Fasciani, Stefano (2024). A listening agent for live control of synthesis parameters using reinforcement learning, Proceedings of International Conference on AI and Musical Creativity. AIMC. doi: 10.5281/zenodo.15110192. Fulltekst i vitenarkiv Vis sammendrag

  This paper presents a novel approach for developing customized autonomous agents for musical improvisation. The model we propose employs reinforcement learning to adaptively control the parameters of a sound synthesizer in response to live audio from a musician. The agent is trained on a corpus of audio files that exemplify the musician’s instrument and stylistic repertoire. During training, the agent listens and learns to imitate the incoming sound according to a set of perceptual descriptors by continuously adjusting the parameters of the synthesizer it controls. To achieve this objective, the agent learns specific strategies that are characteristic of its autonomous behavior in a live interaction. In the paper, we provide a detailed description of the design and implementation of the model for the agent and discuss its application in three selected scenarios.

• Madaghiele, Vincenzo; Lisena, Pasquale & Troncy, Raphaël (2021). MINGUS: Melodic Improvisation Neural Generator Using Seq2Seq, Proceedings of the 22nd International Society for Music Information Retrieval Conference, Online, Nov 7-12, 2021. International Society for Music Information Retrieval. ISSN 9781732729902. doi: 10.5281/zenodo.5625684. Fulltekst i vitenarkiv Vis sammendrag

  Sequence to Sequence (Seq2Seq) approaches have shown good performances in automatic music generation. We introduce MINGUS, a Transformer-based Seq2Seq architecture for modelling and generating monophonic jazz melodic lines. MINGUS relies on two dedicated embedding models (respectively for pitch and duration) and exploits in prediction features such as chords (current and following), bass line, position inside the measure. The obtained results are comparable with the state of the art of music generation with neural models, with particularly good performances on jazz music.

Se alle arbeider i NVA

• Madaghiele, Vincenzo & Kelsey, Cotton, (2025). Recursive speculation. doi: 10.5281/zenodo.17306433. Fulltekst i vitenarkiv Vis sammendrag

  Recursive Speculation is an improvised vocal performance that features a vocalist and the Multi-Agent Autonomous Looper (MAAL), a co-creative sampler/looper based on a multi-agent logic algorithm and machine listening. The MAAL is composed of several agents, each controlling a loop track, which can autonomously decide to sample and play back segments of a live vocal performance by listening to each other. The performance evolves as a dialogue between the musician and the system, in which the machine acts as speculative mirror able to autonomously select and repeat segments of the performer’s voice, layering them to construct evolving musical structures. The system appropriates vocal segments, it disembodies them through repetition, reimagining their meaning and their musical function by recombination and association. The musician and the system are in a co-creative feedback where the musician generates the material the system selects, and at the same time she must respond to the system’s autonomous choices throughout the performance.

• Madaghiele, Vincenzo (2024). Experience Replay. doi: https:/aimc2024.pubpub.org/pub/eghiq141/release/1. Fulltekst i vitenarkiv Vis sammendrag

  Experience Replay is an improvised performance for guitar and live electronics. The sounds generated by the guitar throughout the performance are continuously recorded, re-combined and processed by a system composed of two granular synthesis modules. Machine listening algorithms are used to extract perceptual qualities of the sounds produced by the guitar and the two synthesis modules, generating modulation signals that are used to affect the synthesis parameters. The modulation signals are used to create a feedback loop between the two synthesis units. This generates a complex and unpredictable co-creative system which responds to the guitar, providing an engaging improvisation playground for the performer.

• Madaghiele, Vincenzo (2024). Stutter. doi: https:/www.youtube.com/watch?v=DtuVZ4XHsz0. Fulltekst i vitenarkiv Vis sammendrag

  stutter is a semi-improvised piece for self-playing piano and live electronics. The piano improvises by recombining the score of Miroirs (1905) by Maurice Ravel, while listening to a version of itself altered by live electronic sound processing. The sound evolution of the work is based on a dialogue between two different algorithms, one of which controls the live generation of a piano score and another one controlling the live electronic processing of it. The two systems are coupled in a feedback loop, where the live electronics processes the sound of the piano in real-time, and the score generation is conditioned on the altered electronic version of the piano, instead of the clean sound of it. This technique generates a complex evolving process with which a live musician interacts during the piece.

• Madaghiele, Vincenzo & Demir, Arife Dila (2024). Pain Creature. doi: https:/nime.org/proc_music/nime2024_music_32/index.html. Fulltekst i vitenarkiv Vis sammendrag

  Pain Creature is an interactive wearable instrument for improvisational movement and sound-making. The instrument materializes the distinct qualities of the performer's chronic pain through visual-tactile-auditory aspects. In this semi-improvised dance performance, Pain Creature embodies the dancer's pain as an unwanted presence, a parasite living on her skin that responds to touch with abstract sonic utterances. The piece unfolds as the dancer explores her relationship with the creature, accompanied by a texture of evolving electronic sounds generated by a live electronic musician. As the piece progresses, the performer explores the choreographic and sonic possibilities of the instrument in a dialogue with the electronic sonic background.

Se alle arbeider i NVA