Publications

Journals

  • [Trilla and Alias, 2012] Trilla, A. and Alias, F., "Sentence-based Sentiment Analysis for Expressive Text-to-Speech", In IEEE Transactions on Audio, Speech and Language Processing (ISSN: 1558-7916), September, 2012. [abs] [bib] [link]
    Abstract
    Current research to improve state of the art Text-To- Speech (TTS) synthesis studies both the processing of input text and the ability to render natural expressive speech. Focusing on the former as a front-end task in the production of synthetic speech, this article investigates the proper adaptation of a Sentiment Analysis procedure (positive/neutral/negative) that can then be used as an input feature for expressive speech synthesis. To this end, we evaluate different combinations of textual features and classifiers to determine the most appropriate adaptation procedure. The effectiveness of this scheme for Sentiment Analysis is evaluated using the Semeval 2007 dataset and a Twitter corpus, for their affective nature and their granularity at the sentence level, which is appropriate for an expressive TTS scenario. The experiments conducted validate the proposed procedure with respect to the state of the art for Sentiment Analysis.
    BibTex
    @article{TrillaAlias2012,
    title = {{Sentence-based Sentiment Analysis for Expressive Text-to-Speech}},
    author={Alexandre Trilla and Francesc Al\'{i}as},
    journal={IEEE Trans. Audio, Speech, Lang. Process.},
    year={2012},
    month={Sep.},
    ISSN={1558-7916}
    }

Conferences

2012

  • [Trilla and Alias, 2012] Trilla, A. and Alias, F., "Three-class Sentiment Analysis adapted to short texts", In the XXVIII Conference of the Spanish Society for Natural Language Processing (SEPLN 2012), 2012, September, Castellon, Spain. [abs] [bib] [pdf]
    Abstract
    The demand for information on opinions and sentiment has seen an increase in recent years. This article adapts a general Sentiment Analysis scheme to deal with short texts and three classes of sentiment. It addresses positive, negative and neutral sentiments automatically using Feature Engineering and Text Classification techniques. The effectiveness of this scheme is evaluated using the Semeval 2007 dataset and it achieves maximum rate of 49\%, improving by 7\% the results reported in the state of the art following the same evaluation conditions.
    BibTex
    @misc{TrillaAlias2012,
    author = {Alexandre Trilla and Francesc Al\'{i}as},
    title = {{Three-class Sentiment Analysis adapted to short texts}},
    booktitle = {Procesamiento del Lenguaje Natural},
    year = {2012},
    location = {Castellon, Spain},
    month = {Sep.}
    }
  • [Lozano, et al., 2012] Lozano, I., Trilla, A. and Alias, F., "Spanish JavaSimLib: a tool to compute the semantic similarity between words in Spanish", In the XXVIII Conference of the Spanish Society for Natural Language Processing (SEPLN 2012), 2012, September, Castellon, Spain. (in Spanish) [abs] [bib] [pdf]
    Abstract
    This article adapts JavaSimLib to Spanish, which is a tool to compute the semantic similarity between words according to their information content. To this end, the Spanish WordNet is rearranged into the Lucene search engine, allowing an effective distribution of all the necessary information. The Spanish adaptation procedure that is shown attains the same performance as the original implementation of JavaSimLib in English.
    BibTex
    @misc{Lozanoetal2012,
    author = {Isaac Lozano and Alexandre Trilla and Francesc Al\'{i}as},
    title = {{Spanish JavaSimLib: a tool to compute the semantic similarity between words in Spanish}},
    booktitle = {Procesamiento del Lenguaje Natural},
    year = {2012},
    location = {Castellon, Spain},
    month = {Sep.}
    }
  • [Trilla and Alias, 2012] Trilla, A. and Alias, F., "Sentiment Analysis of Twitter messages based on Multinomial Naive Bayes", In the XXVIII Conference of the Spanish Society for Natural Language Processing (SEPLN 2012) -- Workshop on Sentiment Analysis, 2012, September, Castellon, Spain. [abs] [bib] [pdf]
    Abstract
    This article adapts a Text Classification scheme based on Multinomial Naive Bayes to deal with Twitter messages labelled with six classes of sentiment as well as with their topic. The effectiveness of this scheme is evaluated using the TASS-SEPLN Twitter dataset and it achieves maximum macroaveraged $F_1$ measure rate of 36.28\%.
    BibTex
    @misc{TrillaAlias2012tass,
    author = {Alexandre Trilla and Francesc Al\'{i}as},
    title = {{Sentiment Analysis of Twitter messages based on Multinomial Naive Bayes}},
    booktitle = {Procesamiento del Lenguaje Natural -- Workshop on Sentiment Analysis},
    year = {2012},
    location = {Castellon, Spain},
    month = {Sep.}
    }

2010

  • [Formiga et al., 2010] Formiga, Ll., Trilla, A., Alias, F., Iriondo, I. and Socoro, J.C., "Adaptation of the URL-TTS system to the 2010 Albayzin Evaluation Campaign", In Proceedings of VI Jornadas en Tecnologia del Habla (FALA2010) (ISBN: 978-84-8158-510-0), pp. 363-370, 2010, November, Vigo, Spain. [abs] [bib] [pdf]
    Abstract
    This paper presents the text-to-speech (TTS) synthesis system of La Salle (Universitat Ramon Llull, URL) and its adaptation to the Albayzin Evaluation Campaign of FALA2010 conference. The URL-TTS system follows the classical scheme of unit se- lection TTS synthesis systems. However, it presents two dis- tinguishable particularities: i) prosody prediction learned from labelled data by means of Case-Based-Reasoning (CBR) and perceptual weight tuning by means of active interactive Genetic Algorithms (aiGA). The aiGA-based weights are compared to multilinear regression (MLR) weights both considering classi- cal averaged cost function and its root-mean squared variant. The internal validation tests and the results of the evaluation campaing are described, and finally discussed.
    BibTex
    @inproceedings{ttsChallenge_url_fala2010,
    author = {Llu\'{i}s Formiga and Alexandre Trilla and Francesc Al\'{i}as and Ignasi Iriondo and Joan Claudi Socor\'o},
    title = {{Adaptation of the URL-TTS system to the 2010 Albayzin Evaluation Campaign}},
    booktitle = {Proceedings of VI Jornadas en Tecnolog\'{i}a del Habla (FALA2010)},
    pages = {363--370},
    isbn = {978-84-8158-510-0},
    organization = {Group on Multimedia Technologies (GTM) of the University of Vigo, the Spanish Thematic Network on Speech Technology (RTTH) and the ISCA- Special Interest Group on Iberian Languages (SIG-IL)},
    month = {November},
    address = {Vigo, Spain},
    year = {2010}
    }
  • [Trilla et al., 2010] Trilla, A., Alias, F. and Lozano, I., "Text classification of domain-styled text and sentiment-styled text for expressive speech synthesis", In Proceedings of VI Jornadas en Tecnologia del Habla (FALA2010) (ISBN: 978-84-8158-510-0), pp. 75-78, 2010, November, Vigo, Spain. [abs] [bib] [pdf]
    Abstract
    In the context of text processing for Text-to-Speech (TTS) syn- thesis, this work aims to automatically direct the expressive- ness in speech through tagging the input text appropriately. Since the nature of text presents different characteristics accord- ing to whether it is domain-dependent (related to its topics) or sentiment-dependent, it is studied how these traits influence the identification of expressiveness in text. To this end, two principal Text Classification (TC) methods are considered: a graph-based approach named the Reduced As- sociative Relational Network and the Maximum Entropy clas- sifier. Their effectiveness in domain/sentiment dependent envi- ronments is evaluated. The results indicate that moving from a domain-dependent environment to a more general sentiment- dependent environment strictly results in poorer effectiveness rates, despite the sensible direct association that sentiment pro- vides for dealing with expressiveness. Additionally, it is also evaluated how sensitive the classifiers are to a small increase of training data, yielding a slight positive influence.
    BibTex
    @inproceedings{fala2010,
    author = {Alexandre Trilla and Francesc Al\'{i}as and Isaac Lozano},
    title = {{Text classification of domain-styled text and sentiment-styled text for expressive speech synthesis}},
    booktitle = {Proceedings of VI Jornadas en Tecnolog\'{i}a del Habla (FALA2010)},
    pages = {75--78},
    isbn = {978-84-8158-510-0},
    organization = {Group on Multimedia Technologies (GTM) of the University of Vigo, the Spanish Thematic Network on Speech Technology (RTTH) and the ISCA- Special Interest Group on Iberian Languages (SIG-IL)},
    month = {November},
    address = {Vigo, Spain},
    year = {2010}
    }

2009

  • [Trilla and Alias, 2009] Trilla, A. and Alias, F., "Sentiment classification in English from sentence-level annotations of emotions regarding models of affect", In Proceedings of the 10th Annual Conference of the International Speech Communication Association (Interspeech 2009) (ISSN: 1990-9772), pp. 516-519, 2009, September, Brighton, UK. [abs] [bib] [pdf]
    Abstract
    This paper presents a text classifier for automatically tagging the sentiment of input text according to the emotion that is being conveyed. This system has a pipelined framework composed of Natural Language Processing modules for feature extraction and a hard binary classifier for decision making between positive and negative categories. To do so, the Semeval 2007 dataset composed of sentences emotionally annotated is used for training purposes after being mapped into a model of affect. The resulting scheme stands a first step towards a complete emotion classifier for a future automatic expressive text-to-speech synthesizer.
    BibTex
    @inproceedings{is09ml,
    author = {Alexandre Trilla and Francesc Al\'{i}as},
    title = {Sentiment classification in {E}nglish from sentence-level annotations of emotions regarding models of affect},
    booktitle = {Proceedings of the 10th Annual Conference of the International Speech Communication Association (Interspeech 2009)},
    issn = {1990-9772},
    pages = {516--519},
    organization = {International Speech Communication Association},
    month = {September},
    address = {Vigo, Spain},
    year = {2009}
    }

Dissertations

  • [Trilla, 2010] Trilla, A., "Natural Language Processing techniques applied to speech technologies", Ph.D.(ABD) Thesis, Enginyeria i Arquitectura La Salle, Universitat Ramon Llull, 2010, Barcelona, Spain. [abs] [bib] [pdf]
    Abstract
    This dissertation relates a first research stage in the pursuit of the Ph.D. degree in "Information and communication Technologies and its Management" at La Salle ­ Universitat Ramon Llull. The dissertation presents an overview of the various approaches to Text- based Sentiment Prediction in order to reveal their effectiveness in the tripartite sentiment recognition task, i.e., the identification of positive, negative and neutral orientations in text. It discusses the application of the bases that support these diverse proposals, from the feature extraction stage to the classification phase, contrasting the contributions that each method yields through a set of experiments with the Semeval 2007 dataset and the Fifty Word Fiction corpus.
    BibTex
    @mastersthesis{dea_atrilla,
    author = {Alexandre Trilla},
    title = {{Natural Language Processing techniques applied to speech technologies}},
    school = {{Departament d'Educacio de la Generalitat de Catalunya and Enginyeria i Arquitectura La Salle (Universitat Ramon Llull)}},
    year = {2010},
    address = {Barcelona, Spain},
    }
  • [Trilla, 2008] Trilla, A., "Magnus: Mouse Advanced GNU Speech", Master's Thesis, Enginyeria i Arquitectura La Salle, Universitat Ramon Llull, 2008, Barcelona, Spain. (with honours) [abs] [bib] [pdf]
    Abstract
    This Master's Thesis deals with the development of an application to control the mouse pointer and keyboard arrows of a PC through Catalan voice commands, also documenting the technical aspects involved with the digital signal processing field as well as with the IT field.
    The application is programmed in Java and distributed with four flavors, from the source code distribution aimed at developers to making use of the Java Web Start technology clearly aimed at end users. One of the interesting goals of the project is to provide a degree of oral accessibility for people with reduced mobility.
    BibTex
    @mastersthesis{magnus_master_thesis,
    author = {Alexandre Trilla},
    title = {{M}agnus: {M}ouse {A}dvanced {G}{N}{U} {S}peech},
    school = {{Departament d'Educacio de la Generalitat de Catalunya and Enginyeria i Arquitectura La Salle (Universitat Ramon Llull)}},
    year = {2008},
    address = {{Barcelona, Spain}},
    }
  • [Trilla, 2006] Trilla, A., "Sniffer: el seguidor de linies", Bachelor's Thesis, Enginyeria i Arquitectura La Salle, Universitat Ramon Llull, 2006, Barcelona, Spain. (in Catalan)

Technical reports

  • [Alias, et al., 2011] Alias, F., Socoro, J. C., Iriondo, I., Formiga, L. and Trilla, A., "Application of automatic speech generation techniques for audiovisual production", Quaderns del CAC, n. 37 (ISSN: 1138-9761), December, 2011 (in Catalan) [abs] [bib] [pdf]
    Abstract
    Study on the degree of implementation of speech synthesis technologies in Catalonia in the audiovisual sector from a field that collects the views of their key players and users, such as people with visual disabilities.
    BibTex
    @techreport{cac2011,
    author = {Francesc Al\'{i}as and Joan Claudi Socor\'o and Ignasi Iriondo and Llu\'{i}s Formiga and Alexandre Trilla},
    title = {{Application of automatic speech generation techniques for audiovisual production}},
    institution = {{Consell de l'Audiovisual de Catalunya}},
    year = {2011},
    }

Teaching materials

Discrete-Time Signal Processing

  • Scilab tutorial oriented toward the Practice of Discrete-Time Signal Processing [abs] [bib] [pdf]
    Abstract
    In this tutorial, Scilab is used for signal processing. The several tools needed for completing the Practice of Discrete-Time Signal Pro- cessing are described hereunder. Keep it for reference and use it at your convenience.
    BibTex
    @misc{TrillaSevillanoScilab,
    author = {Alexandre Trilla and Xavier Sevillano},
    title = {{Scilab tutorial oriented toward the Practice of Discrete-Time Signal Processing}},
    howpublished = {(Web-Available)},
    year = {2010}
    }
  • ProcLib - Signal processing library for Scilab [tarball]
  • Filter analysis and design [abs] [bib] [pdf]
    Abstract
    The filter design technique analysed in this tutorial is oriented toward the creation of causal linear-phase Finite Impulse Response (FIR) filters. It is based on the direct approximation of their desired frequency responses by windowing their corresponding impulse responses.
    BibTex
    @misc{TrillaSevillanoFiltdsgn,
    author = {Alexandre Trilla and Xavier Sevillano},
    title = {{Filter analysis and design}},
    howpublished = {(Web-Available)},
    year = {2010}
    }
  • Sampling of continuous-time signals [abs] [bib] [pdf]
    Abstract
    This tutorial presents a thorough description of the sampling procedure of continuous-time signals in order to be processed by a discrete-time signal processing system.
    BibTex
    @misc{TrillaSevillanoSampling,
    author = {Alexandre Trilla and Xavier Sevillano},
    title = {{Sampling of continuous-time signals}},
    howpublished = {(Web-Available)},
    year = {2010}
    }
  • Multirate Discrete-Time Signal Processing [abs] [bib] [pdf]
    Abstract
    This tutorial presents a thorough description of procedure to change the sampling rate using discrete-time processing.
    BibTex
    @misc{TrillaSevillanoMrate,
    author = {Alexandre Trilla and Xavier Sevillano},
    title = {{Multirate Discrete-Time Signal Processing}},
    howpublished = {(Web-Available)},
    year = {2010}
    }
  • Discrete Fourier Transform [abs] [bib] [pdf]
    Abstract
    This tutorial presents the Fourier transform of use in computational environments: the Discrete Fourier Transform.
    BibTex
    @misc{TrillaSevillanoDFT,
    author = {Alexandre Trilla and Xavier Sevillano},
    title = {{Discrete Fourier Transform}},
    howpublished = {(Web-Available)},
    year = {2010}
    }
  • Noise cancellation with adaptive filtering [abs] [bib] [pdf]
    Abstract
    This tutorial presents a practical application of adaptive filtering to cancel noise.
    BibTex
    @misc{TrillaSevillanoNoiseCanc,
    author = {Alexandre Trilla and Xavier Sevillano},
    title = {{Noise cancellation with adaptive filtering}},
    howpublished = {(Web-Available)},
    year = {2011}
    }

Others

  • [Trilla, 2011] Trilla, A., "A skim through HARK, a modular OSS system for robot audition", Working Presentation, 2011, Barcelona, Spain. [abs] [bib] [pdf]
    Abstract
    This working presentation briefly shows the robot audition project HARK, its approaches to noise reduction, its modular framework and its application to dialogue systems.
    BibTex
    @misc{harkpres,
    author = {Alexandre Trilla},
    title = {{A skim through HARK, a modular OSS system for robot audition}},
    month = {Oct.},
    year = {2011},
    }
  • [Trilla, 2009] Trilla, A., "Natural Language Processing techniques in Text-To-Speech synthesis and Automatic Speech Recognition", Working Paper, Enginyeria i Arquitectura La Salle, Universitat Ramon Llull, 2009, Barcelona, Spain. [abs] [bib] [pdf]
    Abstract
    This working paper depicts the usage of Natural Language Processing techniques in the production of voice from an input text, a.k.a. Text-To-Speech synthesis, and the inverse process, which is the production of a written text transcription from an input voice utterance, a.k.a. Automatic Speech Recognition.
    BibTex
    @misc{micnlp09,
    author = {Alexandre Trilla},
    title = {{N}atural {L}anguage {P}rocessing techniques in {T}ext-{T}o-{S}peech synthesis and {A}utomatic {S}peech {R}ecognition},
    month = {January},
    year = {2009},
    }
  • [Trilla, 2002] Trilla, A., "Waiterbot: el robot cambrer", High School Project, Col.legi Episcopal Mare de Deu de l'Academia, 2002, Lleida, Spain. (in Catalan) [pdf] [pic] [pic]

The ramblings of a research engineer on data analysis, software development and innovation management

Contents

Other

Search