Segmentation and Recognition of Multi-Food Meal Images for Carbohydrate Counting

In this paper, we propose novel methodologies for the automatic segmentation and recognition of multi-food images. The proposed methods implement the first modules of a carbohydrate counting and insulin advisory system for type 1 diabetic patients. Initially the plate is segmented using pyramidal mean-shift filtering and a region growing algorithm. Then each of the resulted segments is described by both color and texture features and classified by a support vector machine into one of six different major food classes. Finally, a modified version of the Huang and Dom evaluation index was proposed, addressing the particular needs of the food segmentation problem. The experimental results prove the effectiveness of the proposed method achieving a segmentation accuracy of 88.5% and recognition rate equal to 87%

Automatic Understanding of ATC Speech: Study of Prospectives and Field Experiments for Several Controller Positions

Although there has been a lot of interest in recognizing and understanding air traffic control (ATC) speech, none of the published works have obtained detailed field data results. We have developed a system able to identify the language spoken and recognize and understand sentences in both Spanish and English. We also present field results for several in-tower controller positions. To the best of our knowledge, this is the first time that field ATC speech (not simulated) is captured, processed, and analyzed. The use of stochastic grammars allows variations in the standard phraseology that appear in field data. The robust understanding algorithm developed has 95% concept accuracy from ATC text input. It also allows changes in the presentation order of the concepts and the correction of errors created by the speech recognition engine improving it by 17% and 25%, respectively, absolute in the percentage of fully correctly understood sentences for English and Spanish in relation to the percentages of fully correctly recognized sentences. The analysis of errors due to the spontaneity of the speech and its comparison to read speech is also carried out. A 96% word accuracy for read speech is reduced to 86% word accuracy for field ATC data for Spanish for the "clearances" task confirming that field data is needed to estimate the performance of a system. A literature review and a critical discussion on the possibilities of speech recognition and understanding technology applied to ATC speech are also given.

Improving automatic detection of obstructive sleep apnea through nonlinear analysis of sustained speech

We present a novel approach for the detection of severe obstructive sleep apnea (OSA) based on patients' voices introducing nonlinear measures to describe sustained speech dynamics. Nonlinear features were combined with state-of-the-art speech recognition systems using statistical modeling techniques (Gaussian mixture models, GMMs) over cepstral parameterization (MFCC) for both continuous and sustained speech. Tests were performed on a database including speech records from both severe OSA and control speakers. A 10 % relative reduction in classification error was obtained for sustained speech when combining MFCC-GMM and nonlinear features, and 33 % when fusing nonlinear features with both sustained and continuous MFCC-GMM. Accuracy reached 88.5 % allowing the system to be used in OSA early detection. Tests showed that nonlinear features and MFCCs are lightly correlated on sustained speech, but uncorrelated on continuous speech. Results also suggest the existence of nonlinear effects in OSA patients' voices, which should be found in continuous speech.

Architecture for Text Sign Localization and Recognition

This paper describes a low complexity strategy for detecting and recognizing text signs automatically. Traditional approaches use large image algorithms for detecting the text sign, followed by the application of an Optical Character Recognition (OCR) algorithm in the previously identified areas. This paper proposes a new architecture that applies the OCR to a whole lightly treated image and then carries out the text detection process of the OCR output. The strategy presented in this paper significantly reduces the processing time required for text localization in an image, while guaranteeing a high recognition rate. This strategy will facilitate the incorporation of video processing-based applications into the automatic detection of text sign similar to that of a smartphone. These applications will increase the autonomy of visually impaired people in their daily life.

GMM-based classifiers for the automatic detection of obstructive sleep apnea

The aim of automatic pathological voice detection systems is to serve as tools, to medical specialists, for a more objective, less invasive and improved diagnosis of diseases. In this respect, the gold standard for those system include the usage of a optimized representation of the spectral envelope, either based on cepstral coefficients from the mel-scaled Fourier spectral envelope (Mel-Frequency Cepstral Coefficients) or from an all-pole estimation (Linear Prediction Coding Cepstral Coefficients) forcharacterization, and Gaussian Mixture Models for posterior classification. However, the study of recently proposed GMM-based classifiers as well as Nuisance mitigation techniques, such as those employed in speaker recognition, has not been widely considered inpathology detection labours. The present work aims at testing whether or not the employment of such speaker recognition tools might contribute to improve system performance in pathology detection systems, specifically in the automatic detection of Obstructive Sleep Apnea. The testing procedure employs an Obstructive Sleep Apnea database, in conjunction with GMM-based classifiers looking for a better performance. The results show that an improved performance might be obtained by using such approach.

Detection and pattern recognition applied to leaves and chromosomes

The Project you are about to see it is based on the technologies used on object detection and recognition, especially on leaves and chromosomes. To do so, this document contains the typical parts of a scientific paper, as it is what it is. It is composed by an Abstract, an Introduction, points that have to do with the investigation area, future work, conclusions and references used for the elaboration of the document. The Abstract talks about what are we going to find in this paper, which is technologies employed on pattern detection and recognition for leaves and chromosomes and the jobs that are already made for cataloguing these objects. In the introduction detection and recognition meanings are explained. This is necessary as many papers get confused with these terms, specially the ones talking about chromosomes. Detecting an object is gathering the parts of the image that are useful and eliminating the useless parts. Summarizing, detection would be recognizing the objects borders. When talking about recognition, we are talking about the computers or the machines process, which says what kind of object we are handling. Afterwards we face a compilation of the most used technologies in object detection in general. There are two main groups on this category: Based on derivatives of images and based on ASIFT points. The ones that are based on derivatives of images have in common that convolving them with a previously created matrix does the treatment of them. This is done for detecting borders on the images, which are changes on the intensity of the pixels. Within these technologies we face two groups: Gradian based, which search for maximums and minimums on the pixels intensity as they only use the first derivative. The Laplacian based methods search for zeros on the pixels intensity as they use the second derivative. Depending on the level of details that we want to use on the final result, we will choose one option or the other, because, as its logic, if we used Gradian based methods, the computer will consume less resources and less time as there are less operations, but the quality will be worse. On the other hand, if we use the Laplacian based methods we will need more time and resources as they require more operations, but we will have a much better quality result. After explaining all the derivative based methods, we take a look on the different algorithms that are available for both groups. The other big group of technologies for object recognition is the one based on ASIFT points, which are based on 6 image parameters and compare them with another image taking under consideration these parameters. These methods disadvantage, for our future purposes, is that it is only valid for one single object. So if we are going to recognize two different leaves, even though if they refer to the same specie, we are not going to be able to recognize them with this method. It is important to mention these types of technologies as we are talking about recognition methods in general. At the end of the chapter we can see a comparison with pros and cons of all technologies that are employed. Firstly comparing them separately and then comparing them all together, based on our purposes. Recognition techniques, which are the next chapter, are not really vast as, even though there are general steps for doing object recognition, every single object that has to be recognized has its own method as the are different. This is why there is not a general method that we can specify on this chapter. We now move on into leaf detection techniques on computers. Now we will use the technique explained above based on the image derivatives. Next step will be to turn the leaf into several parameters. Depending on the document that you are referring to, there will be more or less parameters. Some papers recommend to divide the leaf into 3 main features (shape, dent and vein] and doing mathematical operations with them we can get up to 16 secondary features. Next proposition is dividing the leaf into 5 main features (Diameter, physiological length, physiological width, area and perimeter] and from those, extract 12 secondary features. This second alternative is the most used so it is the one that is going to be the reference. Following in to leaf recognition, we are based on a paper that provides a source code that, clicking on both leaf ends, it automatically tells to which specie belongs the leaf that we are trying to recognize. To do so, it only requires having a database. On the tests that have been made by the document, they assure us a 90.312% of accuracy over 320 total tests (32 plants on the database and 10 tests per specie]. Next chapter talks about chromosome detection, where we shall pass the metaphasis plate, where the chromosomes are disorganized, into the karyotype plate, which is the usual view of the 23 chromosomes ordered by number. There are two types of techniques to do this step: the skeletonization process and swiping angles. Skeletonization progress consists on suppressing the inside pixels of the chromosome to just stay with the silhouette. This method is really similar to the ones based on the derivatives of the image but the difference is that it doesnt detect the borders but the interior of the chromosome. Second technique consists of swiping angles from the beginning of the chromosome and, taking under consideration, that on a single chromosome we cannot have more than an X angle, it detects the various regions of the chromosomes. Once the karyotype plate is defined, we continue with chromosome recognition. To do so, there is a technique based on the banding that chromosomes have (grey scale bands] that make them unique. The program then detects the longitudinal axis of the chromosome and reconstructs the band profiles. Then the computer is able to recognize this chromosome. Concerning the future work, we generally have to independent techniques that dont reunite detection and recognition, so our main focus would be to prepare a program that gathers both techniques. On the leaf matter we have seen that, detection and recognition, have a link as both share the option of dividing the leaf into 5 main features. The work that would have to be done is to create an algorithm that linked both methods, as in the program, which recognizes leaves, it has to be clicked both leaf ends so it is not an automatic algorithm. On the chromosome side, we should create an algorithm that searches for the beginning of the chromosome and then start to swipe angles, to later give the parameters to the program that searches for the band profiles. Finally, on the summary, we explain why this type of investigation is needed, and that is because with global warming, lots of species (animals and plants] are beginning to extinguish. That is the reason why a big database, which gathers all the possible species, is needed. For recognizing animal species, we just only have to have the 23 chromosomes. While recognizing a plant, there are several ways of doing it, but the easiest way to input a computer is to scan the leaf of the plant. RESUMEN. El proyecto que se puede ver a continuación trata sobre las tecnologías empleadas en la detección y reconocimiento de objetos, especialmente de hojas y cromosomas. Para ello, este documento contiene las partes típicas de un paper de investigación, puesto que es de lo que se trata. Así, estará compuesto de Abstract, Introducción, diversos puntos que tengan que ver con el área a investigar, trabajo futuro, conclusiones y biografía utilizada para la realización del documento. Así, el Abstract nos cuenta qué vamos a poder encontrar en este paper, que no es ni más ni menos que las tecnologías empleadas en el reconocimiento y detección de patrones en hojas y cromosomas y qué trabajos hay existentes para catalogar a estos objetos. En la introducción se explican los conceptos de qué es la detección y qué es el reconocimiento. Esto es necesario ya que muchos papers científicos, especialmente los que hablan de cromosomas, confunden estos dos términos que no podían ser más sencillos. Por un lado tendríamos la detección del objeto, que sería simplemente coger las partes que nos interesasen de la imagen y eliminar aquellas partes que no nos fueran útiles para un futuro. Resumiendo, sería reconocer los bordes del objeto de estudio. Cuando hablamos de reconocimiento, estamos refiriéndonos al proceso que tiene el ordenador, o la máquina, para decir qué clase de objeto estamos tratando. Seguidamente nos encontramos con un recopilatorio de las tecnologías más utilizadas para la detección de objetos, en general. Aquí nos encontraríamos con dos grandes grupos de tecnologías: Las basadas en las derivadas de imágenes y las basadas en los puntos ASIFT. El grupo de tecnologías basadas en derivadas de imágenes tienen en común que hay que tratar a las imágenes mediante una convolución con una matriz creada previamente. Esto se hace para detectar bordes en las imágenes que son básicamente cambios en la intensidad de los píxeles. Dentro de estas tecnologías nos encontramos con dos grupos: Los basados en gradientes, los cuales buscan máximos y mínimos de intensidad en la imagen puesto que sólo utilizan la primera derivada; y los Laplacianos, los cuales buscan ceros en la intensidad de los píxeles puesto que estos utilizan la segunda derivada de la imagen. Dependiendo del nivel de detalles que queramos utilizar en el resultado final nos decantaremos por un método u otro puesto que, como es lógico, si utilizamos los basados en el gradiente habrá menos operaciones por lo que consumirá más tiempo y recursos pero por la contra tendremos menos calidad de imagen. Y al revés pasa con los Laplacianos, puesto que necesitan más operaciones y recursos pero tendrán un resultado final con mejor calidad. Después de explicar los tipos de operadores que hay, se hace un recorrido explicando los distintos tipos de algoritmos que hay en cada uno de los grupos. El otro gran grupo de tecnologías para el reconocimiento de objetos son los basados en puntos ASIFT, los cuales se basan en 6 parámetros de la imagen y la comparan con otra imagen teniendo en cuenta dichos parámetros. La desventaja de este método, para nuestros propósitos futuros, es que sólo es valido para un objeto en concreto. Por lo que si vamos a reconocer dos hojas diferentes, aunque sean de la misma especie, no vamos a poder reconocerlas mediante este método. Aún así es importante explicar este tipo de tecnologías puesto que estamos hablando de técnicas de reconocimiento en general. Al final del capítulo podremos ver una comparación con los pros y las contras de todas las tecnologías empleadas. Primeramente comparándolas de forma separada y, finalmente, compararemos todos los métodos existentes en base a nuestros propósitos. Las técnicas de reconocimiento, el siguiente apartado, no es muy extenso puesto que, aunque haya pasos generales para el reconocimiento de objetos, cada objeto a reconocer es distinto por lo que no hay un método específico que se pueda generalizar. Pasamos ahora a las técnicas de detección de hojas mediante ordenador. Aquí usaremos la técnica explicada previamente explicada basada en las derivadas de las imágenes. La continuación de este paso sería diseccionar la hoja en diversos parámetros. Dependiendo de la fuente a la que se consulte pueden haber más o menos parámetros. Unos documentos aconsejan dividir la morfología de la hoja en 3 parámetros principales (Forma, Dentina y ramificación] y derivando de dichos parámetros convertirlos a 16 parámetros secundarios. La otra propuesta es dividir la morfología de la hoja en 5 parámetros principales (Diámetro, longitud fisiológica, anchura fisiológica, área y perímetro] y de ahí extraer 12 parámetros secundarios. Esta segunda propuesta es la más utilizada de todas por lo que es la que se utilizará. Pasamos al reconocimiento de hojas, en la cual nos hemos basado en un documento que provee un código fuente que cucando en los dos extremos de la hoja automáticamente nos dice a qué especie pertenece la hoja que estamos intentando reconocer. Para ello sólo hay que formar una base de datos. En los test realizados por el citado documento, nos aseguran que tiene un índice de acierto del 90.312% en 320 test en total (32 plantas insertadas en la base de datos por 10 test que se han realizado por cada una de las especies]. El siguiente apartado trata de la detección de cromosomas, en el cual se debe de pasar de la célula metafásica, donde los cromosomas están desorganizados, al cariotipo, que es como solemos ver los 23 cromosomas de forma ordenada. Hay dos tipos de técnicas para realizar este paso: Por el proceso de esquelotonización y barriendo ángulos. El proceso de esqueletonización consiste en eliminar los píxeles del interior del cromosoma para quedarse con su silueta; Este proceso es similar a los métodos de derivación de los píxeles pero se diferencia en que no detecta bordes si no que detecta el interior de los cromosomas. La segunda técnica consiste en ir barriendo ángulos desde el principio del cromosoma y teniendo en cuenta que un cromosoma no puede doblarse más de X grados detecta las diversas regiones de los cromosomas. Una vez tengamos el cariotipo, se continua con el reconocimiento de cromosomas. Para ello existe una técnica basada en las bandas de blancos y negros que tienen los cromosomas y que son las que los hacen únicos. Para ello el programa detecta los ejes longitudinales del cromosoma y reconstruye los perfiles de las bandas que posee el cromosoma y que lo identifican como único. En cuanto al trabajo que se podría desempeñar en el futuro, tenemos por lo general dos técnicas independientes que no unen la detección con el reconocimiento por lo que se habría de preparar un programa que uniese estas dos técnicas. Respecto a las hojas hemos visto que ambos métodos, detección y reconocimiento, están vinculados debido a que ambos comparten la opinión de dividir las hojas en 5 parámetros principales. El trabajo que habría que realizar sería el de crear un algoritmo que conectase a ambos ya que en el programa de reconocimiento se debe clicar a los dos extremos de la hoja por lo que no es una tarea automática. En cuanto a los cromosomas, se debería de crear un algoritmo que busque el inicio del cromosoma y entonces empiece a barrer ángulos para después poder dárselo al programa que busca los perfiles de bandas de los cromosomas. Finalmente, en el resumen se explica el por qué hace falta este tipo de investigación, esto es que con el calentamiento global, muchas de las especies (tanto animales como plantas] se están empezando a extinguir. Es por ello que se necesitará una base de datos que contemple todas las posibles especies tanto del reino animal como del reino vegetal. Para reconocer a una especie animal, simplemente bastará con tener sus 23 cromosomas; mientras que para reconocer a una especie vegetal, existen diversas formas. Aunque la más sencilla de todas es contar con la hoja de la especie puesto que es el elemento más fácil de escanear e introducir en el ordenador.

New methodology for the integration of biometric features in speaker recognition systems applied to security environments

La cuestión principal abordada en esta tesis doctoral es la mejora de los sistemas biométricos de reconocimiento de personas a partir de la voz, proponiendo el uso de una nueva parametrización, que hemos denominado parametrización biométrica extendida dependiente de género (GDEBP en sus siglas en inglés). No se propone una ruptura completa respecto a los parámetros clásicos sino una nueva forma de utilizarlos y complementarlos. En concreto, proponemos el uso de parámetros diferentes dependiendo del género del locutor, ya que como es bien sabido, la voz masculina y femenina presentan características diferentes que deberán modelarse, por tanto, de diferente manera. Además complementamos los parámetros clásicos utilizados (MFFC extraídos de la señal de voz), con un nuevo conjunto de parámetros extraídos a partir de la deconstrucción de la señal de voz en sus componentes de fuente glótica (más relacionada con el proceso y órganos de fonación y por tanto con características físicas del locutor) y de tracto vocal (más relacionada con la articulación acústica y por tanto con el mensaje emitido). Para verificar la validez de esta propuesta se plantean diversos escenarios, utilizando diferentes bases de datos, para validar que la GDEBP permite generar una descripción más precisa de los locutores que los parámetros MFCC clásicos independientes del género. En concreto se plantean diferentes escenarios de identificación sobre texto restringido y texto independiente utilizando las bases de datos de HESPERIA y ALBAYZIN. El trabajo también se completa con la participación en dos competiciones internacionales de reconocimiento de locutor, NIST SRE (2010 y 2012) y MOBIO 2013. En el primer caso debido a la naturaleza de las bases de datos utilizadas se obtuvieron resultados cercanos al estado del arte, mientras que en el segundo de los casos el sistema presentado obtuvo la mejor tasa de reconocimiento para locutores femeninos. A pesar de que el objetivo principal de esta tesis no es el estudio de sistemas de clasificación, sí ha sido necesario analizar el rendimiento de diferentes sistemas de clasificación, para ver el rendimiento de la parametrización propuesta. En concreto, se ha abordado el uso de sistemas de reconocimiento basados en el paradigma GMM-UBM, supervectores e i-vectors. Los resultados que se presentan confirman que la utilización de características que permitan describir los locutores de manera más precisa es en cierto modo más importante que la elección del sistema de clasificación utilizado por el sistema. En este sentido la parametrización propuesta supone un paso adelante en la mejora de los sistemas de reconocimiento biométrico de personas por la voz, ya que incluso con sistemas de clasificación relativamente simples se consiguen tasas de reconocimiento realmente competitivas. ABSTRACT The main question addressed in this thesis is the improvement of automatic speaker recognition systems, by the introduction of a new front-end module that we have called Gender Dependent Extended Biometric Parameterisation (GDEBP). This front-end do not constitute a complete break with respect to classical parameterisation techniques used in speaker recognition but a new way to obtain these parameters while introducing some complementary ones. Specifically, we propose a gender-dependent parameterisation, since as it is well known male and female voices have different characteristic, and therefore the use of different parameters to model these distinguishing characteristics should provide a better characterisation of speakers. Additionally, we propose the introduction of a new set of biometric parameters extracted from the components which result from the deconstruction of the voice into its glottal source estimate (close related to the phonation process and the involved organs, and therefore the physical characteristics of the speaker) and vocal tract estimate (close related to acoustic articulation and therefore to the spoken message). These biometric parameters constitute a complement to the classical MFCC extracted from the power spectral density of speech as a whole. In order to check the validity of this proposal we establish different practical scenarios, using different databases, so we can conclude that a GDEBP generates a more accurate description of speakers than classical approaches based on gender-independent MFCC. Specifically, we propose scenarios based on text-constrain and text-independent test using HESPERIA and ALBAYZIN databases. This work is also completed with the participation in two international speaker recognition evaluations: NIST SRE (2010 and 2012) and MOBIO 2013, with diverse results. In the first case, due to the nature of the NIST databases, we obtain results closed to state-of-the-art although confirming our hypothesis, whereas in the MOBIO SRE we obtain the best simple system performance for female speakers. Although the study of classification systems is beyond the scope of this thesis, we found it necessary to analise the performance of different classification systems, in order to verify the effect of them on the propose parameterisation. In particular, we have addressed the use of speaker recognition systems based on the GMM-UBM paradigm, supervectors and i-vectors. The presented results confirm that the selection of a set of parameters that allows for a more accurate description of the speakers is as important as the selection of the classification method used by the biometric system. In this sense, the proposed parameterisation constitutes a step forward in improving speaker recognition systems, since even when using relatively simple classification systems, really competitive recognition rates are achieved.

Contributions to Speech Analytics based on Speech Recognition and Topic Identification

La última década ha sido testigo de importantes avances en el campo de la tecnología de reconocimiento de voz. Los sistemas comerciales existentes actualmente poseen la capacidad de reconocer habla continua de múltiples locutores, consiguiendo valores aceptables de error, y sin la necesidad de realizar procedimientos explícitos de adaptación. A pesar del buen momento que vive esta tecnología, el reconocimiento de voz dista de ser un problema resuelto. La mayoría de estos sistemas de reconocimiento se ajustan a dominios particulares y su eficacia depende de manera significativa, entre otros muchos aspectos, de la similitud que exista entre el modelo de lenguaje utilizado y la tarea específica para la cual se está empleando. Esta dependencia cobra aún más importancia en aquellos escenarios en los cuales las propiedades estadísticas del lenguaje varían a lo largo del tiempo, como por ejemplo, en dominios de aplicación que involucren habla espontánea y múltiples temáticas. En los últimos años se ha evidenciado un constante esfuerzo por mejorar los sistemas de reconocimiento para tales dominios. Esto se ha hecho, entre otros muchos enfoques, a través de técnicas automáticas de adaptación. Estas técnicas son aplicadas a sistemas ya existentes, dado que exportar el sistema a una nueva tarea o dominio puede requerir tiempo a la vez que resultar costoso. Las técnicas de adaptación requieren fuentes adicionales de información, y en este sentido, el lenguaje hablado puede aportar algunas de ellas. El habla no sólo transmite un mensaje, también transmite información acerca del contexto en el cual se desarrolla la comunicación hablada (e.g. acerca del tema sobre el cual se está hablando). Por tanto, cuando nos comunicamos a través del habla, es posible identificar los elementos del lenguaje que caracterizan el contexto, y al mismo tiempo, rastrear los cambios que ocurren en estos elementos a lo largo del tiempo. Esta información podría ser capturada y aprovechada por medio de técnicas de recuperación de información (information retrieval) y de aprendizaje de máquina (machine learning). Esto podría permitirnos, dentro del desarrollo de mejores sistemas automáticos de reconocimiento de voz, mejorar la adaptación de modelos del lenguaje a las condiciones del contexto, y por tanto, robustecer al sistema de reconocimiento en dominios con condiciones variables (tales como variaciones potenciales en el vocabulario, el estilo y la temática). En este sentido, la principal contribución de esta Tesis es la propuesta y evaluación de un marco de contextualización motivado por el análisis temático y basado en la adaptación dinámica y no supervisada de modelos de lenguaje para el robustecimiento de un sistema automático de reconocimiento de voz. Esta adaptación toma como base distintos enfoque de los sistemas mencionados (de recuperación de información y aprendizaje de máquina) mediante los cuales buscamos identificar las temáticas sobre las cuales se está hablando en una grabación de audio. Dicha identificación, por lo tanto, permite realizar una adaptación del modelo de lenguaje de acuerdo a las condiciones del contexto. El marco de contextualización propuesto se puede dividir en dos sistemas principales: un sistema de identificación de temática y un sistema de adaptación dinámica de modelos de lenguaje. Esta Tesis puede describirse en detalle desde la perspectiva de las contribuciones particulares realizadas en cada uno de los campos que componen el marco propuesto: _ En lo referente al sistema de identificación de temática, nos hemos enfocado en aportar mejoras a las técnicas de pre-procesamiento de documentos, asimismo en contribuir a la definición de criterios más robustos para la selección de index-terms. – La eficiencia de los sistemas basados tanto en técnicas de recuperación de información como en técnicas de aprendizaje de máquina, y específicamente de aquellos sistemas que particularizan en la tarea de identificación de temática, depende, en gran medida, de los mecanismos de preprocesamiento que se aplican a los documentos. Entre las múltiples operaciones que hacen parte de un esquema de preprocesamiento, la selección adecuada de los términos de indexado (index-terms) es crucial para establecer relaciones semánticas y conceptuales entre los términos y los documentos. Este proceso también puede verse afectado, o bien por una mala elección de stopwords, o bien por la falta de precisión en la definición de reglas de lematización. En este sentido, en este trabajo comparamos y evaluamos diferentes criterios para el preprocesamiento de los documentos, así como también distintas estrategias para la selección de los index-terms. Esto nos permite no sólo reducir el tamaño de la estructura de indexación, sino también mejorar el proceso de identificación de temática. – Uno de los aspectos más importantes en cuanto al rendimiento de los sistemas de identificación de temática es la asignación de diferentes pesos a los términos de acuerdo a su contribución al contenido del documento. En este trabajo evaluamos y proponemos enfoques alternativos a los esquemas tradicionales de ponderado de términos (tales como tf-idf ) que nos permitan mejorar la especificidad de los términos, así como también discriminar mejor las temáticas de los documentos. _ Respecto a la adaptación dinámica de modelos de lenguaje, hemos dividimos el proceso de contextualización en varios pasos. – Para la generación de modelos de lenguaje basados en temática, proponemos dos tipos de enfoques: un enfoque supervisado y un enfoque no supervisado. En el primero de ellos nos basamos en las etiquetas de temática que originalmente acompañan a los documentos del corpus que empleamos. A partir de estas, agrupamos los documentos que forman parte de la misma temática y generamos modelos de lenguaje a partir de dichos grupos. Sin embargo, uno de los objetivos que se persigue en esta Tesis es evaluar si el uso de estas etiquetas para la generación de modelos es óptimo en términos del rendimiento del reconocedor. Por esta razón, nosotros proponemos un segundo enfoque, un enfoque no supervisado, en el cual el objetivo es agrupar, automáticamente, los documentos en clusters temáticos, basándonos en la similaridad semántica existente entre los documentos. Por medio de enfoques de agrupamiento conseguimos mejorar la cohesión conceptual y semántica en cada uno de los clusters, lo que a su vez nos permitió refinar los modelos de lenguaje basados en temática y mejorar el rendimiento del sistema de reconocimiento. – Desarrollamos diversas estrategias para generar un modelo de lenguaje dependiente del contexto. Nuestro objetivo es que este modelo refleje el contexto semántico del habla, i.e. las temáticas más relevantes que se están discutiendo. Este modelo es generado por medio de la interpolación lineal entre aquellos modelos de lenguaje basados en temática que estén relacionados con las temáticas más relevantes. La estimación de los pesos de interpolación está basada principalmente en el resultado del proceso de identificación de temática. – Finalmente, proponemos una metodología para la adaptación dinámica de un modelo de lenguaje general. El proceso de adaptación tiene en cuenta no sólo al modelo dependiente del contexto sino también a la información entregada por el proceso de identificación de temática. El esquema usado para la adaptación es una interpolación lineal entre el modelo general y el modelo dependiente de contexto. Estudiamos también diferentes enfoques para determinar los pesos de interpolación entre ambos modelos. Una vez definida la base teórica de nuestro marco de contextualización, proponemos su aplicación dentro de un sistema automático de reconocimiento de voz. Para esto, nos enfocamos en dos aspectos: la contextualización de los modelos de lenguaje empleados por el sistema y la incorporación de información semántica en el proceso de adaptación basado en temática. En esta Tesis proponemos un marco experimental basado en una arquitectura de reconocimiento en ‘dos etapas’. En la primera etapa, empleamos sistemas basados en técnicas de recuperación de información y aprendizaje de máquina para identificar las temáticas sobre las cuales se habla en una transcripción de un segmento de audio. Esta transcripción es generada por el sistema de reconocimiento empleando un modelo de lenguaje general. De acuerdo con la relevancia de las temáticas que han sido identificadas, se lleva a cabo la adaptación dinámica del modelo de lenguaje. En la segunda etapa de la arquitectura de reconocimiento, usamos este modelo adaptado para realizar de nuevo el reconocimiento del segmento de audio. Para determinar los beneficios del marco de trabajo propuesto, llevamos a cabo la evaluación de cada uno de los sistemas principales previamente mencionados. Esta evaluación es realizada sobre discursos en el dominio de la política usando la base de datos EPPS (European Parliamentary Plenary Sessions - Sesiones Plenarias del Parlamento Europeo) del proyecto europeo TC-STAR. Analizamos distintas métricas acerca del rendimiento de los sistemas y evaluamos las mejoras propuestas con respecto a los sistemas de referencia. ABSTRACT The last decade has witnessed major advances in speech recognition technology. Today’s commercial systems are able to recognize continuous speech from numerous speakers, with acceptable levels of error and without the need for an explicit adaptation procedure. Despite this progress, speech recognition is far from being a solved problem. Most of these systems are adjusted to a particular domain and their efficacy depends significantly, among many other aspects, on the similarity between the language model used and the task that is being addressed. This dependence is even more important in scenarios where the statistical properties of the language fluctuates throughout the time, for example, in application domains involving spontaneous and multitopic speech. Over the last years there has been an increasing effort in enhancing the speech recognition systems for such domains. This has been done, among other approaches, by means of techniques of automatic adaptation. These techniques are applied to the existing systems, specially since exporting the system to a new task or domain may be both time-consuming and expensive. Adaptation techniques require additional sources of information, and the spoken language could provide some of them. It must be considered that speech not only conveys a message, it also provides information on the context in which the spoken communication takes place (e.g. on the subject on which it is being talked about). Therefore, when we communicate through speech, it could be feasible to identify the elements of the language that characterize the context, and at the same time, to track the changes that occur in those elements over time. This information can be extracted and exploited through techniques of information retrieval and machine learning. This allows us, within the development of more robust speech recognition systems, to enhance the adaptation of language models to the conditions of the context, thus strengthening the recognition system for domains under changing conditions (such as potential variations in vocabulary, style and topic). In this sense, the main contribution of this Thesis is the proposal and evaluation of a framework of topic-motivated contextualization based on the dynamic and non-supervised adaptation of language models for the enhancement of an automatic speech recognition system. This adaptation is based on an combined approach (from the perspective of both information retrieval and machine learning fields) whereby we identify the topics that are being discussed in an audio recording. The topic identification, therefore, enables the system to perform an adaptation of the language model according to the contextual conditions. The proposed framework can be divided in two major systems: a topic identification system and a dynamic language model adaptation system. This Thesis can be outlined from the perspective of the particular contributions made in each of the fields that composes the proposed framework: _ Regarding the topic identification system, we have focused on the enhancement of the document preprocessing techniques in addition to contributing in the definition of more robust criteria for the selection of index-terms. – Within both information retrieval and machine learning based approaches, the efficiency of topic identification systems, depends, to a large extent, on the mechanisms of preprocessing applied to the documents. Among the many operations that encloses the preprocessing procedures, an adequate selection of index-terms is critical to establish conceptual and semantic relationships between terms and documents. This process might also be weakened by a poor choice of stopwords or lack of precision in defining stemming rules. In this regard we compare and evaluate different criteria for preprocessing the documents, as well as for improving the selection of the index-terms. This allows us to not only reduce the size of the indexing structure but also to strengthen the topic identification process. – One of the most crucial aspects, in relation to the performance of topic identification systems, is to assign different weights to different terms depending on their contribution to the content of the document. In this sense we evaluate and propose alternative approaches to traditional weighting schemes (such as tf-idf ) that allow us to improve the specificity of terms, and to better identify the topics that are related to documents. _ Regarding the dynamic language model adaptation, we divide the contextualization process into different steps. – We propose supervised and unsupervised approaches for the generation of topic-based language models. The first of them is intended to generate topic-based language models by grouping the documents, in the training set, according to the original topic labels of the corpus. Nevertheless, a goal of this Thesis is to evaluate whether or not the use of these labels to generate language models is optimal in terms of recognition accuracy. For this reason, we propose a second approach, an unsupervised one, in which the objective is to group the data in the training set into automatic topic clusters based on the semantic similarity between the documents. By means of clustering approaches we expect to obtain a more cohesive association of the documents that are related by similar concepts, thus improving the coverage of the topic-based language models and enhancing the performance of the recognition system. – We develop various strategies in order to create a context-dependent language model. Our aim is that this model reflects the semantic context of the current utterance, i.e. the most relevant topics that are being discussed. This model is generated by means of a linear interpolation between the topic-based language models related to the most relevant topics. The estimation of the interpolation weights is based mainly on the outcome of the topic identification process. – Finally, we propose a methodology for the dynamic adaptation of a background language model. The adaptation process takes into account the context-dependent model as well as the information provided by the topic identification process. The scheme used for the adaptation is a linear interpolation between the background model and the context-dependent one. We also study different approaches to determine the interpolation weights used in this adaptation scheme. Once we defined the basis of our topic-motivated contextualization framework, we propose its application into an automatic speech recognition system. We focus on two aspects: the contextualization of the language models used by the system, and the incorporation of semantic-related information into a topic-based adaptation process. To achieve this, we propose an experimental framework based in ‘a two stages’ recognition architecture. In the first stage of the architecture, Information Retrieval and Machine Learning techniques are used to identify the topics in a transcription of an audio segment. This transcription is generated by the recognition system using a background language model. According to the confidence on the topics that have been identified, the dynamic language model adaptation is carried out. In the second stage of the recognition architecture, an adapted language model is used to re-decode the utterance. To test the benefits of the proposed framework, we carry out the evaluation of each of the major systems aforementioned. The evaluation is conducted on speeches of political domain using the EPPS (European Parliamentary Plenary Sessions) database from the European TC-STAR project. We analyse several performance metrics that allow us to compare the improvements of the proposed systems against the baseline ones.

Positron-emission tomography imaging of long-term shape recognition challenges

Long-term visual memory performance was impaired by two types of challenges: a diazepam challenge on acquisition and a sensory challenge on recognition. Using positron-emission tomography regional cerebral blood flow imaging, we studied the effect of these challenges on regional brain activation during the delayed recognition of abstract visual shapes as compared with a baseline fixation task. Both challenges induced a significant decrease in differential activation in the left fusiform gyrus, suggesting that this region is involved in the automatic or volitional comparison of incoming and stored stimuli. In contrast, thalamic differential activation increased in response to memory challenges. This increase might reflect enhanced retrieval attempts as a compensatory mechanism for restoring recognition performance.

State of the art in continuous speech recognition.

In the past decade, tremendous advances in the state of the art of automatic speech recognition by machine have taken place. A reduction in the word error rate by more than a factor of 5 and an increase in recognition speeds by several orders of magnitude (brought about by a combination of faster recognition search algorithms and more powerful computers), have combined to make high-accuracy, speaker-independent, continuous speech recognition for large vocabularies possible in real time, on off-the-shelf workstations, without the aid of special hardware. These advances promise to make speech recognition technology readily available to the general public. This paper focuses on the speech recognition advances made through better speech modeling techniques, chiefly through more accurate mathematical modeling of speech sounds.

Toward the ultimate synthesis/recognition system.

This paper predicts speech synthesis, speech recognition, and speaker recognition technology for the year 2001, and it describes the most important research problems to be solved in order to arrive at these ultimate synthesis and recognition systems. The problems for speech synthesis include natural and intelligible voice production, prosody control based on meaning, capability of controlling synthesized voice quality and choosing individual speaking style, multilingual and multidialectal synthesis, choice of application-oriented speaking styles, capability of adding emotion, and synthesis from concepts. The problems for speech recognition include robust recognition against speech variations, adaptation/normalization to variations due to environmental conditions and speakers, automatic knowledge acquisition for acoustic and linguistic modeling, spontaneous speech recognition, naturalness and ease of human-machine interaction, and recognition of emotion. The problems for speaker recognition are similar to those for speech recognition. The research topics related to all these techniques include the use of articulatory and perceptual constraints and evaluation methods for measuring the quality of technology and systems.

A comparative evaluation of term recognition algorithms

Automatic Term Recognition (ATR) is a fundamental processing step preceding more complex tasks such as semantic search and ontology learning. From a large number of methodologies available in the literature only a few are able to handle both single and multi-word terms. In this paper we present a comparison of five such algorithms and propose a combined approach using a voting mechanism. We evaluated the six approaches using two different corpora and show how the voting algorithm performs best on one corpus (a collection of texts from Wikipedia) and less well using the Genia corpus (a standard life science corpus). This indicates that choice and design of corpus has a major impact on the evaluation of term recognition algorithms. Our experiments also showed that single-word terms can be equally important and occupy a fairly large proportion in certain domains. As a result, algorithms that ignore single-word terms may cause problems to tasks built on top of ATR. Effective ATR systems also need to take into account both the unstructured text and the structured aspects and this means information extraction techniques need to be integrated into the term recognition process.

Character recognition systems using Fourier transformation in incoherent light

Many attempts have been made to overcome problems involved in character recognition which have resulted in the manufacture of character reading machines. An investigation into a new approach to character recognition is described. Features for recognition are Fourier coefficients. These are generated optically by convolving characters with periodic gratings. The development of hardware to enable automatic measurement of contrast and position of periodic shadows produced by the convolution is described. Fourier coefficients of character sets were measured, many of which are tabulated. Their analysis revealed that a few low frequency sampling points could be selected to recognise sets of numerals. Limited treatment is given to show the effect of type face variations on the values of coefficients which culminated in the location of six sampling frequencies used as features to recognise numerals in two type fonts. Finally, the construction of two character recognition machines is compared and contrasted. The first is a pilot plant based on a test bed optical Fourier analyser, while the second is a more streamlined machine d(3signed for high speed reading. Reasons to indicate that the latter machine would be the most suitable to adapt for industrial and commercial applications are discussed.

Common Scientific Lexicon for Automatic Discourse Analysis of Scientific and Technical Texts

The paper reports on preliminary results of an ongoing research aiming at development of an automatic procedure for recognition of discourse-compositional structure of scientific and technical texts, which is required in many NLP applications. The procedure exploits as discourse markers various domain-independent words and expressions that are specific for scientific and technical texts and organize scientific discourse. The paper discusses features of scientific discourse and common scientific lexicon comprising such words and expressions. Methodological issues of development of a computer dictionary for common scientific lexicon are concerned; basic principles of its organization are described as well. Main steps of the discourse-analyzing procedure based on the dictionary and surface syntactical analysis are pointed out.