Papers by Hani Camille Yehia
HAL (Le Centre pour la Communication Scientifique Directe), Apr 23, 2012
Resumo-Determinar como o modelo de loudness BS.1770 para a radiodifusão poderia incluir medidas d... more Resumo-Determinar como o modelo de loudness BS.1770 para a radiodifusão poderia incluir medidas de objetos sonoros com localizações dinâmicas,é uma questão de estudo em andamento no ITU-R. Este trabalho buscou avaliar o efeito da reverberação na sensação de loudness provocada em participantes de um experimento de casamento de loudness entre pares de estímulos sonoros. Das respostas dos participantes, derivou-se uma curva de ajuste de ganho para a filtragem psicoacústica do modelo ITU-R. As predições de ambos os modelos foram então comparadas por suas proximidades em relação aos intervalos de confiança dos ajustes de nível sonoro feitos pelos participantes.
The problem of synthesizing hyperspectral images from RGB images is ill posed, with potentially i... more The problem of synthesizing hyperspectral images from RGB images is ill posed, with potentially infinite solutions, as it involves estimating data in a high-dimensional space, associated with hyperspectral bands, from limited information in a three-dimensional RGB space. However, under certain conditions related to lighting and physical properties of natural scenes, a feasible solution can be found. This study evaluates four methods for estimating hyperspectral data from RGB images: Ridge linear regression, Minibatch K-means followed by linear regression, a multilayer perceptron (MLP) neural network, and Minibatch K-means combined with an MLP neural network. The results of each method are compared with each other and with the NTIRE 2020 Challenge. The comparison was performed using the Mean Absolute Relative Error (MARE) and execution time. The MLP method attained the lowest MARE (0.072) but with the longest execution time (220 s). Ridge regression attained the shortest execution ti...
BMC Neuroscience, 2021
Background A common problem in neurophysiological signal processing is the extraction of meaningf... more Background A common problem in neurophysiological signal processing is the extraction of meaningful information from high dimension, low sample size data (HDLSS). We present RoLDSIS (regression on low-dimension spanned input space), a regression technique based on dimensionality reduction that constrains the solution to the subspace spanned by the available observations. This avoids regularization parameters in the regression procedure, as needed in shrinkage regression methods. Results We applied RoLDSIS to the EEG data collected in a phonemic identification experiment. In the experiment, morphed syllables in the continuum /da/–/ta/ were presented as acoustic stimuli to the participants and the event-related potentials (ERP) were recorded and then represented as a set of features in the time-frequency domain via the discrete wavelet transform. Each set of stimuli was chosen from a preliminary identification task executed by the participant. Physical and psychophysical attributes we...
Music Perception, 2018
This article seeks to unveil quantitative relations between the patterns of movement recurrence o... more This article seeks to unveil quantitative relations between the patterns of movement recurrence of a group of expert clarinetists and expressive sonic manipulations they employ during their performances. The main hypothesis is that the recurrent ancillary gestures of musicians are closely related to their sounded expressive intentions, and that the expressive content imposed by them according to the music structure is reflected in their movement patterns. To conduct this multimodal investigation of expressiveness in music, movement and audio analyses of several clarinet performances of excerpts in the classical repertoire are presented and discussed in conjunction. The results show strong correlations between the recurrence pattern of clarinetists’ ancillary movements and expressive manipulations of timing, timbre, and loudness associated with melodic phrasing and harmonic and dynamic transitions in the performed music excerpts.
A series of studies have shown that musicians accompany themselves better than others. This is kn... more A series of studies have shown that musicians accompany themselves better than others. This is known as the self/other effect in action recognition and simulation and has been demonstrated in other action-perception tasks like recognizing handwriting and the outcomes of dart throwing. In this study, patterns of synchronization and performance coupling were analyzed among professional clarinetists while performing in a simulated orchestral environment. GLMM models ( G eneralized Linear Mixed Model ) applied to acoustic parameters related to the musician’s expressive intentions demonstrate a higher degree of coupling in performances where musicians accompanied themselves compared to when they accompanied other musicians., as well as an improvement in performance when accompanying other musicians as experiments repeated, suggesting musicians were capable of learning synchronization and performance coupling.
2012 IEEE International Conference on Green Computing and Communications, 2012
ABSTRACT
Mechanomyography is a technique that measures the mechanical signal observed from the muscle surf... more Mechanomyography is a technique that measures the mechanical signal observed from the muscle surface during the contraction. It is expected to be a promising tool to evaluate uterine dynamics even though the uterus contraction is known to be involuntary. This study aimed to describe the development of a portable device based on mechanomyography to monitor uterine contraction. To assess the system was performed isometric contractions of the rectus femoris muscle with varying loads. The acquired data showed that the system could identify each contraction's onset, offset, and amplitude. The sensors and the system architecture proved to be a promising tool for the desired application.
The problem of synthesizing hyperspectral images from RGB images is ill posed, with potentially i... more The problem of synthesizing hyperspectral images from RGB images is ill posed, with potentially infinite solutions, as it involves estimating data in a high-dimensional space, associated with hyperspectral bands, from limited information in a three-dimensional RGB space. However, under certain conditions related to lighting and physical properties of natural scenes, a feasible solution can be found. This study evaluates four methods for estimating hyperspectral data from RGB images: Ridge linear regression, Minibatch K-means followed by linear regression, a multilayer perceptron (MLP) neural network, and Minibatch K-means combined with an MLP neural network. The results of each method are compared with each other and with the NTIRE 2020 Challenge. The comparison was performed using the Mean Absolute Relative Error (MARE) and execution time. The MLP method attained the lowest MARE (0.072) but with the longest execution time (220 s). Ridge regression attained the shortest execution ti...
XXVIII Encontro da SOBRAC, 2018
A fadiga mental e a sonolência são reconhecidas como causas determinantes de acidentes e incident... more A fadiga mental e a sonolência são reconhecidas como causas determinantes de acidentes e incidentes relacionados ao erro humano na aviação. No Brasil, de acordo com o CENIPA (Centro de Investigação e Prevenção de Acidentes Aeronáuticos), o aumento de acidentes no modal aéreo entre 2006 e 2013 foi de 158%, atingindo uma taxa de 1 acidente a cada 2 dias, sendo que os fatores humanos estavam presentes em 90% dos casos. Este artigo descreve um estudo retrospectivo da comunicação oral entre piloto e torre de controle de tráfego aéreo em um caso de acidente e teve como objetivo a detecção da fadiga e sonolência por meio da análise de voz e fala. A análise detalhada do acidente realizada pelo CENIPA indicou que a sonolência e a fadiga provavelmente contribuíram para o acidente. Os dados da voz e da fala foram analisados em duas situações: dados registrados trinta e cinco horas antes do acidente aéreo (condição não sonolenta), que foram comparados com amostras coletadas cerca de 1 hora antes do acidente e também durante o acidente (condição de sono). As gravações foram analisadas acusticamente para a extração de medidas objetivas da organização temporal do discurso como: hesitações, pausas silenciosas, prolongamento de sílabas finais e taxa de articulação. Os resultados sugerem que os métodos adotados neste estudo são viáveis para a detecção de fadiga e sonolência por meio da análise de fala. A avaliação da fala é apontada como um método não invasivo promissor para a detecção de fadiga e sonolência em operações da vida real, como o transporte aéreo. No entanto, são necessários estudos de validação em grande escala para determinar se os resultados obtidos no presente estudo de caso podem ser replicados em outros indivíduos e para outras línguas.
Anais de XXXIII Simpósio Brasileiro de Telecomunicações, 2015
Resumo-O comportamento de correlatos acústicosà soprosidade vocal foi investigado em diferentes n... more Resumo-O comportamento de correlatos acústicosà soprosidade vocal foi investigado em diferentes níveis de compressão. O codec utilizado foi o IETF Opus com bitrates entre 16 e 256 kbps. As medidas escolhidas foram: Smoothed cepstral peak prominence (CPPS), spectral flatness of the residue signal (SFRS), pitch amplitude (PA) e spectrographic signal-to-noise-ratio (S 2 NR). A investigação foi conduzida em duas etapas: inicialmente usando amostras de voz sintética que tinham como referência a relação sinal-ruído controlada e posteriormente amostras de voz natural classificadas perceptivamente em soprosidade. As medidas acústicas mantiveram alta correlação com as respectivas referências: SNR para voz sintética e soprosidade para voz natural. O erro quadrático médio estabilizou-se na maior parte das medidas até a taxa de 128 kbps. Considerando os resultados, o Opus não comprometeu a avaliação da soprosidade e sua adoção irá permitir economia em armazenamento e largura de banda na transmissão das amostras.
Aerospace Medicine and Human Performance, 2019
atigue can, from a behavioral perspective, be defined as the state of performance impairment foll... more atigue can, from a behavioral perspective, be defined as the state of performance impairment following a period of mental or physical effort. 9 Sleepiness refers to the transition between wakefulness and sleep and is defined as the tendency to fall asleep even when the individual is supposed to be active, for example, performing work. However, both fatigue and sleepiness are characterized by decreased ability to work, to operate machinery safely, attention failures, cognitive slowing, and memory problems. 2,7 The etiological factors most commonly cited for fatigue are intense physical activity, prolonged or excessive exertion, insufficient sleep, prolonged wakefulness, and physical and emotional illnesses. 7,12 In addition, several work-related and environmental factors in an aircraft can cause or contribute to fatigue in pilots. These include high mental workload, shift scheduling characteristics, reduced cockpit space, diversified maneuvers, alternating acceleration forces, poor airflow, low luminosity, and continuous background noise and vibration. The consequences of factors contributing to fatigue and sleepiness are likely to be potentiated when the individual is, at the same time, carrying out safety-critical activities that demand sustained attention and concentration. 1,4 It is emphasized that fatigue and sleepiness are often used as synonyms, but they differ because sleepiness is mitigated by sleep, while fatigue mitigation demands rest. 5 Speech evaluation has been suggested as a promising method for detection of fatigue and sleepiness in real life operations,
Revista Brasileira de Ensino de Física, 2016
Este artigo apresenta uma análise sob a ótica de teoria da informação de algumas maneiras de cria... more Este artigo apresenta uma análise sob a ótica de teoria da informação de algumas maneiras de criar senhas para dificultar um ataque por força bruta. A lei de Zipf é observada nas línguas naturais e, por conseguinte, a entropia é reduzida quando as utilizamos ao criar uma senha. Muitas das empresas utilizam a política de restringir o número de caracteres de uma senha. Além disso, queremos criar senhas que não sejam demasiadas longas, nem que sejam difíceis de se memorizar. Mostraremos que, dado este cenário, a melhor estratégia (relação de compromisso entre criar uma senha forte e uma senha de fácil memorização e utilização) é a utilização de acrônimos, senhas formadas pela combinação (aparentemente sem nexo) da primeira letra de palavras em uma frase. Com esta abordagem podemos aumentar em aproximadamente 80% a entropia por caractere.
In order to map the spectral characteristics of the great variety of sounds a single musical inst... more In order to map the spectral characteristics of the great variety of sounds a single musical instrument may produce, different notes were performed and sampled in several intensity levels across the whole extension of a clarinet. Amplitude and frequency timevarying curves of partials were measured by Discrete Fourier Transform. A limited set of orthogonal spectral bases was derived by Principal Component Analysis techniques. These bases defined spectral subspaces capable of representing all tested sounds and of grouping of them, which were validated by Mean Opinion Score auditory tests of similarity. Subspaces involving larger groups of notes were used to compare the sounds according to the distance metrics of the representation.
cefala.org
Page 1. ESTUDO DA VARIAÇÃO DO TIMBRE DA CLARINETA NA INTERPRETAÇÃO DE FRASES MUSICAIS ATRAVÉS DE ... more Page 1. ESTUDO DA VARIAÇÃO DO TIMBRE DA CLARINETA NA INTERPRETAÇÃO DE FRASES MUSICAIS ATRAVÉS DE ANÁLISE POR COMPONENTES PRINCIPAIS DA DISTRIBUIÇÃO ESPECTRAL Hugo B. de Paula ...
Resumo O objetivo deste trabalho é analisar e verificar as limitações de técnicas de mascaramento... more Resumo O objetivo deste trabalho é analisar e verificar as limitações de técnicas de mascaramento binário no domínio tempo-freqüência para o problema de separação cega de fontes (Blind Source Separation) a partir de misturas reverberadas. Tais técnicas baseiam-se na esparsidade dos sinais de voz e consideram que, em uma mistura de fontes independentes, é improvável encontrar sinais ativos em um mesmo instante de tempo e freqüência. Neste trabalho, inicialmente, dado que encontrar máscaras ideais com base ...
IEIC Technical Report (Institute of Electronics, Information and Communication Engineers), 1999
Abstract; Previously, we reported a system for animating talking heads based on principal compone... more Abstract; Previously, we reported a system for animating talking heads based on principal component analysis (PCA) for a small set of static 3D faces, and subsequently driven by meansures of 3D motion for a small set of locations on the face. Though PCA afforded massive reduction in the 3D kinematic data needed to represent talking faces, the resulting animations were not always adequate for all speech postures. In this paper we present improved results based on use of larger sets of gestural postures, including emotional and ...
Uploads
Papers by Hani Camille Yehia