1,721,204 research outputs found
A new Pitch Tracking Smoother based on Deep Neural Networks.
No full textThis paper presents a new pitch tracking smoother based on deep neural networks (DNN). The proposed system has been extensively tested using two reference benchmarks for English and exhib- ited very good performances in correcting pitch detection algorithms outputs
Neural Models for the Automatic Processing of Italian
No full textThe volume reports the author’s research experiences and experiments in developing solutions in the various areas of Computational Linguistics and Natural Language Processing. The book focuses mainly on applications based on Deep Neural Networks, but contrasting these approaches with the methodologies used in the past, and it is organised in such a way as to both describe the state of the art in this discipline, examining the studies proposed by the author, and to outline a useful path also for the training of young scholars and students. Following the same spirit, the volume can be profitably read both by people more concentrated on humanistic studies and people with more technical interests
On Automatic Prominence Detection for German
Full text linkTamburini F, Wagner P. On Automatic Prominence Detection for German. In: Proceedings of Interspeech 2007. 2007: 1809-1812.Perceptual prominence is an important indicator of a word's and syllable's lexical, syntactic,
semantic and pragmatic status in a discourse. Its automatic annotation would be a valuable
enrichment of large databases used in unit selection speech synthesis and speech recognition.
While much research has been carried out on the interaction between prominence and
acoustic factors, little progress has been made in its automatic annotation. Previous
approaches to German relied on linguistic features in prominence detection, but a purely
acoustic method would be advantageous. We applied an algorithm to German data that had
been previously used for English and Italian. Both the algorithm and the data annotation
encode prominence as a continuous rather than a categorical parameter. First results are
encouraging, but again show that prominence perception relies on linguistic expectancies as
well as acoustic patterns. Also, our results further strengthen the view that force accents are a
more reliable cue to prominence than pitch accents in German
Objective, Subjective and Linguistic Roads to Perceptual Prominence. How are they compared and why?
Full text linkWagner P, Tamburini F, Windmann A. Objective, Subjective and Linguistic Roads to Perceptual Prominence. How are they compared and why? In: Proceedings of Interspeech 2012. 2012: 2386-2389
Electra-AGE_FE
No full textThis is a new system for Frame Identification (FI), based on pre-trained text encoders trained discriminatively and graphs embedding, producing state of the art performance and. We take in consideration all the extremely different procedures used to evaluate systems for this task performing a complete evaluation over two benchmarks and all possible splits and cleaning procedures used in the FI literature
A dynamic model for reference corpora structure definition
No full textA representative corpus of written Italian - CORIS - constructed at the Centre for Theoretical and Applied Linguistics of Bologna University (CILTA) is available on-line. Considering the importance of the comparability of reference corpora in interlinguistic studies, a further corpus - CODIS - was designed. Aimed at specialist needs, CODIS presents a dynamic and adaptive structure providing for the selection of the subcorpora pertinent to a specific research project and allowing the researcher to define the size of each subcorpus. CODIS is designed to be dynamically adapted by the scholar to different comparative needs by a careful combination of small corpus chunks of various types and sizes. The chunk sizes were carefully selected in order to allow for various combinations creating subcorpora of different sizes, ranging from 0 to the maximum size of each CORIS subcorpus. This fine granularity provides a wide range of corpora composition options, satisfying almost all comparative needs
Automatic detection of prosodic prominence in continuous speech
No full textThis paper presents work in progress on the automatic detection of prosodic prominence in continuous speech. Prosodic prominence involves two different phonetic features: pitch accents, connected with fundamental frequency (F0) movements and syllable overall energy, and stress, which exhibits a strong correlation with syllable duration and high-frequency emphasis. By deriving a set of acoustic parameters it is possible to build syllable-stress detectors as well as pitch-accent detectors and combine them to build an automatic system devoted to prominence detection. Starting from a syllable-segmented utterance, the system presented here is capable of correctly identify prominent syllables with an agreement, with human-tagged data, comparable with the inter-human agreement reported in the literature
CORISTagger (versione 1.0)
No full textCORISTagger è un programma per l’annotazione automatica di testi in lingua italiana rispetto alle categorie lessicali. E’ in grado di associare ad ogni termine contenuto nel testo la sua parte del discorso risolvendo le ambiguità con un livello di precisione estremamente elevato, anche grazie all’incorporazione di un analizzatore morfologico altamente sofisticato basato su un lemmario composto da circa 120.000 lemmi.
L’annotatore ha ottenuto risultati eccellenti nella recente campagna di valutazione EVALITA2007 classificandosi tra i migliori sistemi per la lingua italiana e raggiungendo prestazioni allo stato dell'arte nel settore.
Questo annotatore è stato utilizzato con successo per annotare il corpus CORIS/CODIS, corpus di riferimento per l'italiano contemporaneo.
CORISTagger è il risultato di uno studio condotto dall'autore negli ultimi anni. I risultati sono già stati presentati a convegni e su rivista:
* Tamburini F. (2000). Annotazione grammaticale e lemmatizzazione di corpora in italiano, Linguistica e informatica: multimedialita', corpora e percorsi di apprendimento, Rossini Favretti R. (a cura di), Bulzoni, Roma, 57-73.
* Tamburini F. (2007). CORISTagger: a high-performance PoS tagger for Italian. Intelligenza Artificiale, IV(2), 14-15
QWSD
No full textThis is a novel algorithm for Word Sense Disambiguation (WSD) based on Quantum Probability Theory. The Quantum WSD algorithm requires concepts representations as vectors in the complex domain and thus we have developed a technique for computing com- plex word and sentence embeddings based on the Paragraph Vectors algorithm. De- spite the proposed method is quite simple and that it does not require long training phases, when it is evaluated on a standardized benchmark for this task it exhibits state-of-the-art (SOTA) performances
Prosodic prominence detection in speech
No full textThis paper presents work in progress on the automatic detection of prosodic prominence in continuous speech. Prosodic prominence involves two different phonetic features: pitch accents, connected with fundamental frequency (FO) movements and syllable overall energy, and stress, which exhibits a strong correlation with syllable nuclei duration and high-frequency emphasis. By measuring these acoustic parameters it is possible to build an automatic system capable of correctly identifying prominent syllables with an agreement with human-tagged data comparable with the inter-human agreement reported in the literature. These results were achieved without using any information apart from acoustic parameters. © 2003 IEEE
- …
