The awareness of the powerful human auditory capabilities led to several research initiatives that focused on the benefits of using sound and music as a primary interface for transmitting information. Starting from the second half of 20th century, much international effort has been put in the development of methods act at studying and exploiting this idea. The first official definition of Auditory Display and, in particular, of Sonification in 1992 formalized this concept, marking the beginning of a new era for data transmission techniques, involving an extensive list of interdisciplinary fields. Particular attention has been paid to the study of the benefits that such an approach can give in the domain of sports and motor rehabilitation. The use of real-time auditory feedback of physiological and physical information based on sound signals has been proven particularly useful to motivate, monitor, and modify those processes. Several applications were developed in this field (such as running and cycling), with state-of-theart technologies specifically dealing with activity-synchronized auditory feedback. For this scope, it was necessary to explore several sensors setups combined with time-domain Digital Signal Processing techniques and mathematical models for synchronization in order to retrieve essential information on the spatio-temporal motion data. This approach relies on the exploitation of particular discrete cues present in the input signals, obtained by a meaningful choice of physical variables (e.g., pressure, acceleration) to track the activity rhythm. The positive research results in these contexts opened the door to the possibility of widening the field of study to other domains, for which signals with different time-series characteristics are involved. In this thesis, we propose a new frequency domain approach for synchronized sonification of auditory bio-feedback, in the domain of physiological activities, motor kinematic, and kinetic processes. This study derived from the need to find an valid way to monitor, analyze, and extract useful parameters (e.g., frequency and phase information) from periodic signals for which this task involves issues when executed in the time domain. This effort gave the possibility of studying the benefits of sonification to signals that do not present evident discrete cues in the time series and are characterized by a low frequency range. This goal was achieved involving a combination of different frequency-domain techniques used in signal processing and data analysis, and adapting mathematical models for synchronization already employed in state-of-the-art systems, such as the Kuramoto Model, to our methodology. The outcome of our methodology evaluation provided promising results in this direction, setting a new baseline for future research and developments.
La consapevolezza delle potenzialità dell’apparato uditivo umano ha stimolato numerosi studi incentrati sui vantaggi derivanti dall’utilizzo del suono e della musica come strumento per la trasmissione di informazioni. A partire dalla seconda metà del XX secolo, la comunità scientifica si è impegnata nello sviluppo di metodi atti ad approfondire e sfruttare questa idea. La definizione ufficiale di Auditory Display e, in particolare, di Sonificazione risalente al 1992 ha formalizzato questa metodologia, segnando l’inizio di una nuova era per quanto riguarda lo sviluppo di tecniche di trasmissione di dati in diversi ambiti interdisciplinari. Particolare attenzione è stata dedicata allo studio dei benefici che tale approccio può portare in campo sportivo e riabilitativo. L’uso di feedback audio in risposta ad informazioni fisiologiche e fisiche si è dimostrato utile per motivare, monitorare e modificare tali processi. Diversi sistemi sono stati sviluppati in questi ambiti (ad es., nella corsa e nel ciclismo), grazie all’utilizzo di metodi per lo sviluppo di strategie di feedback audio. Nel corso degli anni sono state testate diverse configurazioni di sensori, tecniche di Digital Signal Processing nel dominio del tempo e modelli matematici per lo studio di fenomeni di sincronizzazione, al fine di ottenere in tempo reale i parametri necessari per l’analisi dell’attività in questione. Tale approccio si basa sullo sfruttamento di particolari elementi discreti presenti nei segnali di input, ottenuti attraverso una scelta significativa di variabili fisiche (ad es., pressione ed accelerazione). I risultati positivi degli studi eseguiti in questi contesti hanno aperto le porte alla possibilità di ampliare il campo di applicazione ad altri settori, per i quali sono coinvolti segnali con caratteristiche temporali differenti. In questa tesi, proponiamo un nuovo approccio nel dominio delle frequenze per la sonificazione sincronizzata di attività e processi fisiologici, cinetici e cinematici. Questo studio nasce dalla necessità di sviluppare un algoritmo efficiente per monitorare, analizzare ed estrarre parametri utili (ad es. frequenza e fase) da segnali periodici per i quali questo compito comporta diverse problematiche se eseguito nel dominio del tempo. Ciò ha dato, inoltre, la possibilità di studiare i benefici della sonificazione applicata a segnali che non presentano eventi discreti nelle serie temporali e coinvolgono una banda di frequenze estremamente bassa. L’ obiettivo è stato raggiunto sfruttando una combinazione di strumenti che operano nel dominio della frequenza utilizzati nell’elaborazione dei segnali e nell’analisi di dati, e adattando modelli matematici per la sincronizzazione già impiegati in passato, come il modello di Kuramoto. I risultati forniti dai nostri esperimenti sono promettenti, e stabiliscono un nuovo punto di partenza per ricerche e sviluppi futuri.
Real-time periodicity analysis of low-frequency and low-energy transients signals. Application to gesture-based sonification systems
VAGHI, ANDREA
2018/2019
Abstract
The awareness of the powerful human auditory capabilities led to several research initiatives that focused on the benefits of using sound and music as a primary interface for transmitting information. Starting from the second half of 20th century, much international effort has been put in the development of methods act at studying and exploiting this idea. The first official definition of Auditory Display and, in particular, of Sonification in 1992 formalized this concept, marking the beginning of a new era for data transmission techniques, involving an extensive list of interdisciplinary fields. Particular attention has been paid to the study of the benefits that such an approach can give in the domain of sports and motor rehabilitation. The use of real-time auditory feedback of physiological and physical information based on sound signals has been proven particularly useful to motivate, monitor, and modify those processes. Several applications were developed in this field (such as running and cycling), with state-of-theart technologies specifically dealing with activity-synchronized auditory feedback. For this scope, it was necessary to explore several sensors setups combined with time-domain Digital Signal Processing techniques and mathematical models for synchronization in order to retrieve essential information on the spatio-temporal motion data. This approach relies on the exploitation of particular discrete cues present in the input signals, obtained by a meaningful choice of physical variables (e.g., pressure, acceleration) to track the activity rhythm. The positive research results in these contexts opened the door to the possibility of widening the field of study to other domains, for which signals with different time-series characteristics are involved. In this thesis, we propose a new frequency domain approach for synchronized sonification of auditory bio-feedback, in the domain of physiological activities, motor kinematic, and kinetic processes. This study derived from the need to find an valid way to monitor, analyze, and extract useful parameters (e.g., frequency and phase information) from periodic signals for which this task involves issues when executed in the time domain. This effort gave the possibility of studying the benefits of sonification to signals that do not present evident discrete cues in the time series and are characterized by a low frequency range. This goal was achieved involving a combination of different frequency-domain techniques used in signal processing and data analysis, and adapting mathematical models for synchronization already employed in state-of-the-art systems, such as the Kuramoto Model, to our methodology. The outcome of our methodology evaluation provided promising results in this direction, setting a new baseline for future research and developments.| File | Dimensione | Formato | |
|---|---|---|---|
|
Andrea_Vaghi_877710_MASTER_THESIS.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
4.31 MB
Formato
Adobe PDF
|
4.31 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/149881