Download Computational Processing of the Portuguese Language: 11th by Jorge Baptista PDF

By Jorge Baptista

ISBN-10: 3319097601

ISBN-13: 9783319097602

ISBN-10: 331909761X

ISBN-13: 9783319097619

This publication constitutes the refereed court cases of the eleventh foreign Workshop on Computational Processing of the Portuguese Language, PROPOR 2014, held in Sao Carlos, Brazil, in October 2014. The 14 complete papers and 19 brief papers provided during this quantity have been conscientiously reviewed and chosen from sixty three submissions. The papers are prepared in topical sections named: speech language processing and purposes; linguistic description, syntax and parsing; ontologies, semantics and lexicography; corpora and language assets and ordinary language processing, instruments and applications.

Show description

Read or Download Computational Processing of the Portuguese Language: 11th International Conference, PROPOR 2014, São Carlos/SP, Brazil, October 6-8, 2014. Proceedings PDF

Similar data mining books

Advanced Methods for Knowledge Discovery from Complex Data

This booklet brings jointly learn articles by means of energetic practitioners and best researchers reporting contemporary advances within the box of data discovery. an summary of the sphere, taking a look at the problems and demanding situations concerned is through insurance of contemporary developments in facts mining. this offers the context for the following chapters on tools and purposes.

Crowdsourcing Geographic Knowledge: Volunteered Geographic Information (VGI) in Theory and Practice

The phenomenon of volunteered geographic details is a part of a profound transformation in how geographic facts, info, and information are produced and circulated. by way of situating volunteered geographic info (VGI) within the context of big-data deluge and the data-intensive inquiry, the 20 chapters during this e-book discover either the theories and functions of crowdsourcing for geographic wisdom construction with 3 sections targeting 1).

Big data Related Technologies, Challenges and Future Prospects

This Springer short offers a accomplished evaluate of the history and up to date advancements of massive information. the price chain of huge info is split into 4 levels: information new release, information acquisition, information garage and information research. for every part, the e-book introduces the final history, discusses technical demanding situations and studies the newest advances.

Extra resources for Computational Processing of the Portuguese Language: 11th International Conference, PROPOR 2014, São Carlos/SP, Brazil, October 6-8, 2014. Proceedings

Sample text

Building a GMM Supervector. A GMM supervector is constructed by stacking the means of the adapted mixture components. 2 "Smile" Features The “Smile” set, derived from the openSMILE toolkit [27], corresponds to 68 measures (“functionals”) applied to a vector of 39 MFCC parameters (statics and dynamics MFCC coefficients). The “functionals” can be divided into six groups: Extremes, Moments, Percentiles, Times, Means and Peaks. Extremes contains 5 parameters (maximum, minimum, range, mean frames between maximum values and mean of frames between minimum values); Moments contains 4 parameters (variance, Characterizing Parkinson's Disease Speech by Acoustic and Phonetic Features 31 skewness, kurtosis and standard deviation); Percentiles contains 9 parameters (3 quartiles, 3 inter-quartile ranges, percentiles 1 and 99 and inter-percentile ranges), Times contains 10 parameters (4 up and down-level times, raise and fall times), Means contains 11 parameters (including arithmetic mean, arithmetic mean of absolute values, quadratic mean among others) and Peaks contains 29 parameters (including number of peaks, mean of frames between peaks, standard deviation of frames between peaks among others).

The performance is undoubtedly boosted by the complete lack of test data from speakers aged 11- 24. 4 ASR Experiments This section describes the ASR experiments carried out to test the potential benefits of automatic age group classification in ASR. We used three different sets of Hidden Markov Model (HMM) models for the experiments: “standard” AMs trained using young to middle-aged adults’ speech (hereafter “SAM”), as well as two separate sets of AMs specifically optimised for children’s and elderly people’s speech (hereafter “CAM” and “EAM”), respectively.

For instance, for Brazilian Portuguese (BP) - standard pronunciation of São Paulo should the verbal form ‘he/she sleeps’ be pronounced as [d'ɔɾmɪ] or [d'oɾmɪ]1? And the form ‘I required’, should it be pronounced as [xek'ɛɾɪ] or [xek'eɾɪ]? For standard European Portuguese (EP), the verbal form 1 For phonetic annotation, we follow International Phonetic Alphabet (IPA) transcription symbols. As default, the examples within this paper are for Brazilian Portuguese, São Paulo standard pronunciation (see section 3 for a brief description of the subject).

Download PDF sample

Computational Processing of the Portuguese Language: 11th International Conference, PROPOR 2014, São Carlos/SP, Brazil, October 6-8, 2014. Proceedings by Jorge Baptista


by Daniel
4.1

Rated 4.55 of 5 – based on 31 votes