This seminar, on the 30th of March 2022, will have two PHD presentations: Judy NAJNUDEL, S3AM team and Claire RICHARDS, PDS team.
If you can't be in the Salle Stravinsky, this link on IRCAM Channel YouTube is for you: https://youtu.be/cvr-QwW079Q
You can review it at the MEDIAS IRCAM site:
Judy Najnudel, SEAM team, direction Thomas Hélie (CNRS - UMR STMS) and co-direction Rémy Muller (UVI):
Physical modeling of nonlinear components for audio effects production.
We are interested in the modeling of nonlinear electronic components for the simulation of analog audio circuits. Among these circuits, we focus in particular on the first generations of compressors, equalizers, and amplifiers used in music production. Our goal is to propose realistic models of components with unknown laws, yet simple enough to allow real-time simulation.
To achieve this goal, we explore two different approaches, based on the Port Hamiltonian Systems (PHS) formalism. Indeed, this formalism preserves the passivity and the power balance of the system, which, coupled with ad hoc numerical methods, guarantees the stability of the simulations.
The first approach is "white box" oriented: the topology of the circuit is assumed to be known, and we focus on the multiphysics modeling of specific components, in particular ferromagnetic coils (present in guitar amplifiers and wah-wah pedals), and photocouplers (present in optical compressors).
The second approach is "grey box" oriented: we try to find the topology of the circuit and the constitutive laws of the components in a concomitant way, using measurements. To do so, we propose to inform the learning of the circuit by an underlying SHP structure, and to deal with the nonlinearities by means of reproducing kernels. In this way, we impose some essential physical properties, while allowing a wide range of nonlinear behaviors. Finally, we try to generalize this type of approach for the treatment of complex circuits, through the introduction of the Koopman operator.
Claire Richards, PDS team, direction Nicolas Misdarris (IRCAM - UMR STMS), and co-direction Roland Cahen (CRD Ensci - Les Ateliers) and industrial partner Actronika (CIFRE):
Designing for multi-modal interactions: Between hearing and touch
This research focuses on two areas of study: design research and perceptual sciences. Research through design is an investigative approach that allows the researcher to formalize their methodologies based on their own creative practice. My practice concerns the creation of audio-haptic devices and the evaluation of how they can stimulate our senses - particularly hearing and touch.
By applying these devices in experimental contexts, I aim to explore several hypotheses. First, the perception of sound by extra-tympanic conduction (often called bone conduction) is not limited to the skull. This mode of sound perception depends on the characteristics of the transmitted vibratory signal, as well as on the position of the stimulus on the body. Finally, when perceived in parallel, extra-tympanic conduction and tactile perception of the vibratory stimulus can contribute to novel multi-modal effects.
Each audio-haptic device formalizes an experimental objective. The first device, a stimulation module that incorporates just one vibratory motor, is designed for a psychophysical study. I asked participants, "Did you hear a signal?" after the presentation of vibratory stimuli to the clavicle, spine, and sternum: areas of the body with musculoskeletal structures close to the surface, which provide pathways for the signal to reach the inner ear. The results allow an estimation of the impact of haptic sensation on the perception of an auditory event.
The second audio-haptic device formalizes another objective: that of exploring effect design for the sense of touch, the sense of hearing and the grey area between the two. This device integrates nine vibratory motors in a wearable harness, targeting the spine, ribs and clavicles. I am preparing experiments whose results will contribute to the definition of an interface for creating multi-sensory effects. The resulting compositions could illustrate potential applications, or simply demonstrate an untapped sensory experience.