Back Six PhD thesis defenses in a period of two week at the MTG

Six PhD thesis defenses in a period of two week at the MTG

14.01.2016

 

In a period of two weeks there are six defenses of PhD thesis carried out at the MTG !!!. They are:

Friday, January 22nd 2016 at 15:30h in room 55.410 (Tanger Building, UPF Communication Campus)
Josep M Comajuncosas: “Assessing creativity in computer music ensembles: a computational approach”
Thesis director: Sergi Jordà; Defense Jury: Xavier Serra (UPF), Enric Guaus (ESMUC) and Atau Tanaka (Goldsmith Univ.)

Abstract: 
Over the last decade Laptop Orchestras and Mobile Ensembles have proliferated. As a result, a large body of research has arisen on infrastructure, evaluation, design principles and compositional methodologies for Computer Music Ensembles (CME).
However, little has been addressed and very little is known about the challenges and opportunities provided by CMEs for creativity in musical performance. Therefore, one of the most common issues CMEs have to deal with is the lack of a systematic approach to handle the implications of the performative paradigms they seek to explore, in terms of their creative constraints and affordances. This is the challenging goal this thesis addresses, and for attaining so it first seeks to find a common ground in the strategies developed for assessing creativity in different performative setups, for later proposing an informed pathway for performative engagement in CMEs.
Our research combines an exploratory stage and an experimental stage. The exploratory stage was informed by out artistic praxis with our own CME, the Barcelona Laptop Orchestra. Through the study of the multi-user instruments developed over the past years, we identified the creative constraints and affordances provided by different performative paradigms. Informed by the findings provided by our artistic research, the experimental stage addressed the study of musical creativity through the performance analysis on specifically designed multi-user instruments. For such purpose we proposed a novel computational methodology to evaluate the creative content of a musical performance.
Two experiments were conducted to incorporate our computational methodology into ecologically valid scenarios, aimed at a better understanding of the relationship between topologies of interdependence and creative outcome. For both experiments, we captured performance data from ensemble improvisations, from where the creativity metrics were then computed. As a preliminary step, we investigated the performative engagement and sharing of musical ideations in an ensemble scenario. In a further step, we computed the creativity attributes to comparatively evaluate performances under different scenarios.
The findings provided quantitative evidence of the differences between musical creativity in individual, ensemble and interdependent scenarios. Additionally, the findings point out what strategies performers adopt to best keep their own musical voice in interdependence scenarios, and what novel creative behaviors may be promoted through new topologies of interdependence. Our findings shed light on the nature of performers’ creative behavior with interdependent multi-user instruments, and show that the introduced methodology can have applications in the broader context of analysis of creativity in musical performance.
 

Friday, January 29th 2016, at 11h in room 55.309 (Tanger Building, UPF Communication Campus)
Martí Umbert: “Expression Control of Singing Voice Synthesis: Modeling Pitch and Dynamics with Unit Selection and Statistical Approaches”
Thesis directors: Jordi Bonada and Xavier Serra; Defense Jury: Rafael Ramírez (UPF), Josep Lluís Arcos (CSIC-IIIA) and Roberto Bresin (KTH).

Abstract:
Sound synthesis technologies have been applied to speech, instruments, and singing voice. While these technologies need to have a sound representation as realistic as possible, the sound synthesis should also reproduce the expressive characteristics of the original sound. This, we refer to emotional speech synthesis, expressive performances of synthesized instruments, as well as expression in singing voice synthesis. Indeed, the singing voice has some commonalities with both speech (the sound source is the same) and instruments (concerning musical aspects such as melody and expression resources). 
Modeling singing voice expression is a difficult task. We are completely familiarized with the singing voice instrument, and thus we easily detect whether artifcially achieved results are similar to a real singer or not. There are many features that should be controlled related to melody, dynamics, rhythm, and timbre, which make achieving natural expression a complex task.
This thesis focuses on the control of a singing voice synthesizer to achieve natural expression similar to a real singer. In this thesis we examine the control of pitch and dynamics. In the unit selection-based system we define the cost functions for unit selection as well as the unit transformations and concatenation steps. The statistically-based systems model both sequences of notes and sequences of note transitions and sustains. Finally, we also present a system which combines the previous ones. These systems are trained with two expression databases that we have designed, recorded, and labeled. These databases comprise sequences of three notes or rests.
Our perceptual evaluation compares the proposed systems with a baseline expression system and a performance-driven approach. The perceptual evaluation shows that the hybrid systems achieves the closest natural expression to a human voice. In the objective evaluation we focus on the systems efficiency.
This thesis delivers numerous contributions to the field of our research: 1) it provides a discussion on expression and summarizes some expression definitions, 2) it reviews previous works on expression control in singing voice synthesis, 3) it provides an online compilation of sound excerpts from different works, 4) it proposes a methodology for expression database creation, 5) it implements a unit selection-based system for expression control, 6) it proposes two statistical-based systems, 7) it presents a hybrid system, 8) it compares the proposed systems with other state of the art systems, 9) it proposes another use case in which the proposed systems can be applied, 10) it provides a set of proposals to improve the evaluation.
 

Monday, February 1st 2016, at 16h in room 55.309 (Tanger Building, UPF Communication Campus)
Panagiotis Papiotis: “A Computational Approach to Studying Interdepence in String Quartet Performance”
Thesis directors: Esteban Maestre and Xavier Serra; Defense Jury: Werner Goebl (Univ. Music and Performing Arts, Vienna), Ralph Andrzejak (UPF) and Josep Lluís Arcos (CSIC)

Abstract: 
This dissertation proposes a computational data-driven methodology to measure music ensemble interdependence - the degree to which musicians interact and influence each other’s actions in order to achieve a shared goal - using a string quartet ensemble as a case study.
We present the outcomes of an experiment on music ensemble interdependence, where we recorded the members of a professional string quartet performing exercises and excerpts of musical pieces in two conditions: solo, where each musician performs their part alone, and ensemble, where the entire quartet performs together following a short rehearsal. During the performance we acquire multimodal data in the form of audio recordings, motion capture data from sound-producing movements and upper body movement, as well as high quality video. All of the recorded data have been published online as an open research dataset.
From the acquired data, we extract numerical features in the form of time series that describe performance in terms of four distinct musical dimensions: intonation, dynamics, timbre, and timing. We apply four different interdependence estimation methods based on time series analysis - Pearson correlation, Mutual Information, Granger causality and Nonlinear Coupling coefficient - to the extracted features in order to assess the overall level of interdependence between the four musicians for each performance dimension individually. We then carry out a statistical comparison of interdependence estimated for the ensemble and solo conditions.
Our results show that it is possible to correctly discriminate between the two experimental conditions for each of the studied performance dimensions. By computing the difference in estimated interdependence between the ensemble and solo condition for a given performance dimension, we are also able to compare across different recordings in terms of the established interdependence and relate the results to the underlying goal of the exercise.
We additionally study the aural perception of music ensemble interdependence, assessing the capability of listeners to distinguish between audio recordings of ensemble performances and artificially synchronized solo performances as a function of the listeners’ own background and the performance dimension that each recording focused on.
The proposed methodology and obtained results explore a novel direction for research on music ensemble interdependence that goes beyond temporal synchronization and towards a broader understanding of joint action in music performance, while the shared dataset provides a valuable resource that serves as a foundation for future studies to build upon.
 

Friday, February 5th 2016, at 11h in room 55.309 (Tanger Building, UPF Communication Campus)
Graham Coleman: “Descriptor Control of Sound Transformations and Mosaicing Synthesis”
Thesis directors: Xavier Serra and Jordi Bonada; Defense Jury: Rarafel Ramírez (UPF), Josep Lluís Arcos (CSIC) and Bob Sturm (QMUL-UK) 

Abstract: 
Sampling, as a musical or synthesis technique, is a way to reuse recorded musical expressions. In this dissertation, several ways to expand sampling synthesis are explored, especially mosaicing synthesis, which imitates target signals by transforming and compositing source sounds, in the manner of a mosaic made of broken tile.
One branch of extension consists of the automatic control of sound transformations towards targets defined in a perceptual space. The approach chosen uses models that predict how the input sound will be transformed as a function of the selected parameters. In one setting, the models are known, and numerical search can be used to find suffcient parameters; in the other, they are unknown and must be learned from data.
Another branch focuses on the sampling itself. By mixing multiple sounds at once, perhaps it is possible to make better imitations, e.g. in terms of the harmony of the target. However, using mixtures leads to new computational problems, especially if properties like continuity, important to high quality sampling synthesis, are to be preserved.
A new mosaicing synthesizer is presented which incorporates all of these elements: supporting automatic control of sound transformations using models, mixtures supported by perceptually relevant harmony and timbre descriptors, and preservation of continuity of the sampling context and transformation parameters. Using listening tests, the proposed hybrid algorithm was compared against classic and contemporary algorithms, and the hybrid algorithm performed well on a variety of quality measures.
 
Friday, February 5th 2016, at 15h in room 55.309 (Tanger Building, UPF Communication Campus)
Stefan Kersten: “Statistical modelling and resynthesis of environmental texture sounds”
Thesis directors: Xavier Serra and Hendrik Purwins; Defense Jury: Rarafel Ramírez (UPF), Enric Guaus (ESMUC) and Diemo Schwarz (IRCAM) 
 
Abstract:
Environmental texture sounds are an integral, though often overlooked, part of our daily life. They constitute those elements of our sounding environment that we tend to perceive subconsciously but which we miss when they are missing. Those sounds are also increasingly important for adding realism to virtual environments, from immersive artificial worlds through computer games to mobile augmented reality systems. This work spans the spectrum from data-driven stochastic sound synthesis methods to distributed virtual reality environments and their aesthetic and technological implications. We propose a framework for statistically modelling environmental texture sounds in different sparse signal representations. We explore three different instantiations of this framework, two of which constitute a novel way of representing texture sounds in a physicallyinspired sparse statistical model and of estimating model parameters from recorded sound examples. We propose a new method of creatively interacting with corpuses of sound segments that are organised in a twodimensional space and evaluate our work in a teaching context for musicians and sound artists. Finally, we describe two different authoring and simulation environments for creating sonic landscapes for virtual reality environments and augmented audio reality applications. These systems serve as a test bed for exploring possible applications of environmental texture sound synthesis models. We evaluate the validity of the developed systems in the context of a prototype virtual reality environment and within the commercial setting of a mobile location based audio platform. We also introduce a novel sound synthesis engine that serves as the basis for realtime rendering of large soundscapes. Its performance is evaluated in the context of a commercial location based audio platform that is in daily use by content producers and end users. In summary, this thesis contributes to the advancement of the state of the art in statistical modelling of environmental sound textures by exploring novel ways of representing those sounds in a sparse setting. Our research also significantly contributed to the succesfull realisation of an innovative location based audio platform.
 

Wednesday, Februrary 10th 2016, at 10am in room 55.309 (Tanger Building, UPF Communication Campus)
Sebastián Mealla: “Designing Sonic Interactions for Implicit Physiological Computing”
Thesis directors: Sergi Jordà and Aleksander Väljamäe; Defense Jury: Rafael Ramírez (UPF), Aureli Soria-Frisch (starlab) and Wendy Ju (Stanford Univ.)

Abstract:
The field of Human-Computer Interaction (HCI) has been historically devoted to understand the interplay between people and computers. However, for the last three decades, it has been mainly based on overt and explicit control by means of peripheral devices such as the keyboard and the mouse. As devices and systems are becoming increasingly complex and powerful, this traditional approach to interface design is often lagging behind, constituting a bottleneck for seamless HCI.
In order to achieve more natural interactions with computer systems, HCI has to go beyond explicit control and incorporate the implicit subtleties of human-human interaction. This could be achieved by means of Physiological Computing, which monitors naturalistic changes in the user psychophysiological states (affective, perceptive or cognitive) for adapting system responses without explicit control. At the output level, Sonic Interaction Design (SID) appears as an excellent medium for representing implicit physiological states, as acoustic data can be processed faster than visual presentation, can be easily localized in space, it has a good temporal resolution, and account for displaying multiple data streams while releasing the visual sense.
Therefore, in this dissertation we aim to conceptualize, prototype and evaluate sonic interaction designs for implicit Physiological Computing in the context of HCI. For achieving this goal, we leverage on physiological sensing techniques, namely EEG and ECG, to estimate user’s implicit states in real time, and apply diverse SID methodologies to adapt system responses according to these statuses. We incrementally develop different implicit sonic interactions (from direct audification to complex musical mappings) and evaluate them in HCI scenarios (from neurofeedback to music performance), assessing their perceptualization quality, the role of mapping complexity, and their meaningfulness in the musical domain.
 

Multimedia

Categories:

SDG - Sustainable Development Goals:

Els ODS a la UPF

Contact