Projects

The Laboratory carries out research projects (funded by public institutions) and knowledge transfer projects (in co-operation with companies).

Knowledge transfer projects

Chiara

Dates: 2016-2018
Funding company: Team Training

Talking Brains

Date: 2016
Funding company: CosmoCaixa

Research projects

Technologies derived from AMPER-CAT and analysis of complementary corpora

PIs: Ana Ma. Fernández-Planas, Lourdes Romera-Barrios
Reference: FFI2015-64859-P (MINECO/FEDER)
Dates: 2016-2020
Website: http://stel3.ub.edu/labfon/amper/cast/ampercat_tecnologias.html

This project has the following goals: 1) completing the study of broad focus statements and information-seeking yes-no questions in the Peninsular and American locales that are missing in the Spanish linguistic domain in the AMPER framework; 2) enriching the corpus with other types of prosodically related phrases such as vocative or wh- questions in the Catalan domain and in selected locales of the Peninsular area in the Spanish linguistic domain; 3) creating technical tools linked to the processing of prosodic data, specifically, an automatic tagger at the phonetic level and at the phonological level according to the AM model and according to the guidelines of the Sp-ToBI and the Cat-ToBI, a program to work in dialectometry with large numerical databases obtained from prosodic phonetic analysis, and an automatic recognizer of geoprosodic characteristics; and 4) enabling a new web page to host the new results. In this way, the study follows the line started in 2003 that is inserted de facto into an international macro project that has its center and international direction at the Université Stendhal-Grenoble III (France) from where a great Multimedia Atlas of Prosody of the Romance Area (AMPER) that includes a wide database of sound, graphics and explanations of the different varieties studied. The methodology that is being followed includes: field work for the elaboration of a digitized corpus, acoustic analysis of the corpus and perceptual verification tests based on the melodic patterns. On the other hand, we intend to go a step further and record other types of sentences (wh- questions and vocatives, for example) with the aim of obtaining a more global knowledge of the prosody of different geographical areas. Finally, we also aim at going beyond the initial objectives of AMPER: we want to develop a more technical line and get three tools initially tightly linked to AMPER and later more generalizable, that can be used to achieve an automatic prosodic labeller at the phonetic and phonological level, a dialectometric analysis program from numerical data and a geoprosodic area recognizer for unknown speakers. Significant results are expected not only for the scientific community, but also for the applied tools that are expected and that will constitute a step forward in one of the applications of phonetics in life in society, the so-called judicial phonetics and phonetics focused on synthesis and automatic recognition.

Geoprosody of Catalan and Spanish in Catalan-speaking areas and of Spanish in Murcia (AMPERCAT): Last locales of controlled speech and assessment of spontaneous speech

PI: Eugenio Martínez Celdrán
Reference: FFI2012-35998
Dates: 2013 - 2016
Website: http://stel.ub.es/labfon/amper

This project constitutes the last phase of an ambitious geolinguistic study of the prosody of declarative and interrogative sentences of Catalan and of the varietis of Spanish spoken in Catalan-Castilian bilingua areas, as well as in Murcia. The general aim is contributing to the creation of a multimedia atlas. This study, which we carried out since 2003, is part of an international project based at the Centre de Dialectologie and the GIPSA-lab of the Université Stendhal-Grenoble III (France). Such Multimedia Atlas includes a wide database of sound, graphics and explanations of the different prosodic varieties of the Romance languages spoken not only in Europe but also in Latin America and the US. The methodology that is being followed includes: field work for the collection of recordings, acoustic analysis of the corpus, and perceptual tests. This project aims at completing the study of prosody in different locales of Catalonia, Aragon (La Franja), Valencia, Balearic Islands and Murcia, as well as the city of L'Alguer, Andorra and the so-called Catalunya Nord (in France).

Towards a geoprosody of Catalan and Spanish in Catalan-speaking areas and of Spanish in Murcia (AMPERCAT)

PI: Eugenio Martínez Celdrán
Reference: FFI2009-09309
Dates: 2009 - 2012
Website: http://stel.ub.es/labfon/amper

This project has been carried out within the AMPER (Multimedia Atlas of Prosody of the Romance Area) framework. AMPER is a European-wide research project on intonation. The main focus is phonetic, with dialectological and sociolinguistic phonological implications. The main objective is to study the intonation of the Romance languages spokien in Europe and to collect the results in maps that are available in internet. The sentences studied include broad focus statements and information-seeking yes-no questions. The research team based in Barcelona 1) coordinates the researches about Catalan, Castilian and Galician in the Iberian Peninsula that are carried out by different local groups, and 2) carries out directly the research about the prosody of Catalan and Spanish spoken in Catalonia.

Multimedia database of Catalan and Spanish in bilingual areas (AMPERCAT)

PI: Eugenio Martínez Celdrán
Reference: HUM2006-05238
Dates: 2006 - 2009

Contribution to the Muiltimedia Atlas of Romance Prosody in Catalonia

PI: Eugenio Martínez Celdrán
Reference: DGICYT - BFF2003 - 08487
Dates: 2003 - 2006

Prosodic features in a model of perception of speech

PI: Eugenio Martínez Celdrán
Reference: DGICYT - PB98 - 1230
Dates: 1999 - 2002

This research project was a continuation of the previous one so that the model had theoretical and practical coherence. In this way, after obtaining the phoneme chain, we set out to study the prosodic features that would allow us to segment the chain and thus complete the automatic recognition program.

Implementation of a model of speech recognition based in acoustic features

PI: Eugenio Martínez Celdrán
Reference: DGICYT - PB94 - 0919
Dates: 1995 - 1998

The objective of this research is to develop a model that represents the way in which the listener processes the acoustic string of speech elements until it is transformed into a phoneme chain. To eliminate an excessive subordination of intuition and inconsistencies of any theoretical model, we have proposed to develop a computational program of automatic speech recognition that simulates the theoretical model using the flow diagram arising from the theoretical model and the invariant indices already investigated in the previous project. The evaluation has been carried out by different male and female speakers.

Speech processing and acoustic invariance in Spanish

PI: Eugenio Martínez Celdrán
Reference: DGICYT - PB91 - 0278
Dates: 1992 - 1994

The objective of this research is to find all the invariant acoustic indices in the consonants and vowels of Spanish. This involves investigating, through different speakers and dialects, those components of the acoustic signal that are presented with the same frequency and amplitude characteristics. The topic has a double interest: first, to develop an acoustic-perceptive phonetics for Spanish, which is almost an unexplored field in the bibliography; secondly, to carry out a work of great importance for the automatic recognition of parliament. The study is inserted in the "theory of acoustic invariance of speech" developed, above all, in England and the United States by researchers such as Blumstein, Stevens, Kewlwy-Port, Repp...