Nelson Neto

An open-source rule-based syllabification tool for Brazilian Portuguese

BackgroundThe automatic syllabification process is an essential prerequisite for speech synthesis systems. However, the task is not trivial, and several techniques have been adopted over the last decade. Furthermore, while there are many public resources for some languages (e.g., English and Japanese), the resources for Brazilian Portuguese (BP) are still limited. This paper discusses ways to diminish this drawback, through the implementation of an open-source syllabification system for BP.MethodsThe proposed tool is based on published rule-based algorithms, with some new proposals, especially in the treatment of words with diphthongs and hiatus.ResultsComputer experiments were performed on a randomly chosen extract of the CETEN-Folha text corpus, and the results showed the percentage of correctly syllabified words of 99%.ConclusionsA subjective evaluation was also conducted in order to compare the elaborated syllabification algorithm with the reference one within a text-to-speech system for BP. All developed codes and databases are publicly available.

Free tools and resources for Brazilian Portuguese speech recognition

An automatic speech recognition system has modules that depend on the language and, while there are many public resources for some languages (e.g., English and Japanese), the resources for Brazilian Portuguese (BP) are still limited. This work describes the development of resources and free tools for BP speech recognition, consisting of text and audio corpora, phonetic dictionary, grapheme-to-phone converter, language and acoustic models. All of them are publicly available and, together with a proposed application programming interface, have been used for the development of several new applications, including a speech module for the OpenOffice suite. Performance tests are presented, comparing the developed BP system with a commercial software. The paper also describes an application that uses synthesis and speech recognition together with a natural language processing module dedicated to statistical machine translation. This application allows the translation of spoken conversations from BP to English and vice versa. The resources make easier the adoption of BP speech technologies by other academic groups and industry.

An open-source speech recognizer for Brazilian Portuguese with a windows programming interface

This work is part of the effort to develop a speech recognition system for Brazilian Portuguese. The resources for the training and test stages of this system, such as corpora, pronunciation dictionary, language and acoustic models, are publicly available. Here, an application programming interface is proposed in order to facilitate using the open-source Julius speech decoder. Performance tests are presented, comparing the developed systems with a commercial software.

Spoltech and OGI-22 Baseline Systems for Speech Recognition in Brazilian Portuguese

Speech processing is a data-driven technology that relies on public corpora and associated resources. In contrast to languages such as English, there are few resources for Brazilian Portuguese (BP). This work describes efforts toward decreasing such gap and presents systems for speech recognition in BP using two public corpora: Spoltech and OGI-22. The following resources are made available: HTK scripts, pronunciation dictionary, language and acoustic models. The work discusses the baselineresults obtained with these resources.

Software usando reconhecimento e síntese de voz: o estado da arte para o Português brasileiro

Speech is a natural interface for human-computer interaction. Speech (or voice) technology is a well-developed field when one considers the international community. There is a wide variety of academic and industrial software. The majority of them assumes a recognizer or synthesizer is available, and can be programmed through an API. In contrast, there are few resources in public domain for Brazilian Portuguese. This work discusses some of these issues and compares SAPI and JSAPI, which are APIs promoted by Microsoft and Sun, respectively. We also present two examples: a tic-tac-toe JSAPI-based game using Portuguese digits recognition and a computer-aided language learning (CALL) application using SAPI-based speech recognition in English and synthesis in Portuguese.

A Proposal of a Universal Remote Control System Based on Head Movements

Technological developments converge to make people interact with electronic devices in an easy way. For people with disabilities, however, that interaction becomes something more than simple: it becomes possible. The current work presents a proposal of an open-source, low cost universal remote control system that translates users head poses into commands to electronic devices. In addition, a proximity sensor circuit was combined to radio-frequency modules in order to act as a wireless switch. Among the applications found with respect to alternative remote controls, none of them supports head gestures as input, which would certainly make them a viable option to people whose upper limbs are compromised. A mean opinion score questionnaire was applied to volunteers in order to evaluate the system. The results show great interest of the users in this kind of technology.

Evaluation of Information Visualization Interaction Techniques Using Gestures and Widgets in 3D Environments

This paper presents the results of usability evaluation of an Information Visualization tool with touchless gestural commands. The tool has well-known visualizations tasks implemented in itself, allowing users to interact on a 3D scatterplot visualization technique. The chosen usability evaluation was the Think Aloud protocol, together with questionnaires conducted by an interviewer, both of them performed with five participants in a controlled environment.

Baseline acoustic models for brazilian portuguese using CMU sphinx tools

Advances in speech processing research rely on the availability of public resources such as corpora, statistical models and baseline systems. In contrast to languages such as English, there are few specific resources for Brazilian Portuguese. This work describes efforts aiming to decrease such gap. Baseline acoustic models for Brazilian Portuguese were built using the CMU Sphinx toolkit and public domain resources: speech corpora, phonetic dictionary and language model. Experiments were carried on for dictation and grammar tasks and the obtained results can be used to support further researches. Part of the trained acoustic models and a reference speech corpus were made publicly available.

Improving Metagenomic Assemblies Through Data Partitioning: A GC Content Approach

Assembling metagenomic data sequenced by NGS platforms poses significant computational challenges, especially due to large volumes of data, sequencing errors, and variations in size, complexity, diversity and abundance of organisms present in a given metagenome. To overcome these problems, this work proposes an open-source, bioinformatic tool called GCSplit, which partitions metagenomic sequences into subsets using a computationally inexpensive metric: the GC content. Experiments performed on real data show that preprocessing short reads with GCSplit prior to assembly reduces memory consumption and generates higher quality results, such as an increase in the size of the largest contig and N50 metric, while both the L50 value and the total number of contigs produced in the assembly were reduced. GCSplit is available at https://github.com/mirand863/gcsplit.

Classification Procedure for Motor Imagery EEG Data.

Brain computer interface establishes a new model of communication, whereby it is possible to communicate using only cerebral signals, that can be obtained from different kind of cerebral stimuli. By the way, one of the most common stimulus is the motor imagery of the arms. However, since a set of variables leads to different levels of classification accuracy, it is necessary to search for procedures that can enhance the recognition accuracy of brain signals in order to create more precise systems. This paper proposes a classification procedure for discrimination of two motor imagery classes obtained using the Emotiv EPOC+ EEG signal acquisition device. The Emotiv EPOC+ has 14 input channels, but only four were used – the ones directly related with the capture of motor imagery signals. The presented procedure was created considering the MI common spatial pattern package from the OpenVibe software and the support vector machine (SVM) classification approach. As well, the procedure runs under the OpenVibe scenarios. A database with motor imagery signals from five subjects was built in order to perform the classification tests. In order to select the best features, several aspects from the signal acquisition until the classification process were analysed, such as selection of the best Kernel to SVM classifier, frequency band, filter output channels, and a grid-search to estimate the classifier parameters. At the end, an increase of 28,96% in the mean accuracy was achieved, regarding to the OpenVibe MI standard scenario.