Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 7
Encontrada(s) 1 página(s)

RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

Evaluación de la revisión de transcripciones y traducciones automáticas de vídeos poliMedia

  • Valor Miró, Juan Daniel
  • Turró Ribalta, Carlos|||0000-0002-3840-9405
  • Civera Saiz, Jorge|||0000-0002-0963-0143
  • Juan, Alfons|||0000-0002-9984-4072
[ES] Los vídeos docentes son una herramienta de gran aceptación en el mundo universitario, que se integran en metodologías activas y innovadoras, dando pie a plataformas como poliMedia, plataforma de la Universitat Politècnica de València (UPV) que permite la creación, publicación y difusión de este tipo de contenido multimedia. En el marco del proyecto europeo transLectures, la UPV ha generado automáticamente transcripciones y traducciones en español, catalán e inglés para todos los vídeos incluidos en el repositorio poliMedia. Las transcripciones y traducciones automáticas generadas tienen una alta calidad. Sin embargo, estas transcripciones y traducciones poseen errores inherentes al proceso automático de subtitulado utilizado, por lo que en ocasiones puede ser necesaria una revisión manual posterior para garantizar la ausencia de errores. El objetivo de este trabajo es evaluar este proceso de revisión manual para poder compararlo en términos de coste temporal con un proceso de obtención completamente manual. Así pues, en el marco de las ayudas de la UPV de Docencia en Red 2013-2014, pusimos en marcha una campaña de evaluación del proceso de revisión de transcripciones y traducciones automáticas por parte del profesorado, cuyos resultados indican inequívocamente la eficacia de estas técnicas y el importante ahorro de tiempo que suponen., The research leading to these results has received funding from the European Union FP7/2007- 2013 under grant agreement no 287755 (transLectures) and ICT PSP/2007-2013 under grant agreement no 621030 (EMMA), and the Spanish MINECO Active2Trans (TIN2012-31723) research project.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

TransLectures

  • Silvestre Cerdà, Joan Albert|||0000-0003-2291-8296
  • Garcés Díaz-Munío, Gonçal|||0000-0002-2594-5858
  • Giménez Pastor, Adrián
  • Civera Saiz, Jorge|||0000-0002-0963-0143
  • Sanchis Navarro, José Alberto|||0000-0002-2943-0990
  • Juan, Alfons|||0000-0002-9984-4072
  • Del Agua Teba, Miguel Angel
  • Gascó Mora, Guillem
  • Martínez-Villaronga, Adrià Agustí
  • Pérez González de Martos, Alejandro Manuel
  • Sánchez-Cortina, Isaías
  • Serrano Martínez-Santos, Nicolás
  • Spencer, Rachel Nadine
  • Valor Miró, Juan Daniel
  • Andrés Ferrer, Jesús
transLectures (Transcription and Translation of Video Lectures) is an EU STREP project in which advanced automatic speech recognition and machine translation techniques are being tested on large video lecture repositories. The project began in November 2011 and will run for three years. This paper will outline the project¿s main motivation and objectives, and give a brief description of the two main repositories being considered: VideoLectures.NET and poliMedia. The first results obtained by the UPV group for the poliMedia repository will also be provided., The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 287755. Funding was also provided by the Spanish Government (iTrans2 project, TIN2009-14511; FPI scholarship BES-2010-033005; FPU scholarship AP2010-4349)
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

Does more data always yield better translations?

  • Gascó Mora, Guillem
  • Rocha Sánchez, Martha Alicia
  • Sanchis Trilles, Germán
  • Andrés Ferrer, Jesús
  • Casacuberta Nolla, Francisco|||0000-0002-8497-5598
Nowadays, there are large amounts of data available to train statistical machine translation systems. However, it is not clear whether all the training data actually help or not. A system trained on a subset of such huge bilingual corpora might outperform the use of all the bilingual data. This paper studies such issues by analysing two training data selection techniques: one based on approximating the probability of an indomain corpus; and another based on infrequent n-gram occurrence. Experimental results not only report significant improvements over random sentence selection but also an improvement over a system trained with the whole available data. Surprisingly, the improvements are obtained with just a small fraction of the data that accounts for less than 0.5% of the sentences. Afterwards, we show that a much larger room for improvement exists, although this is done under non-realistic conditions., The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement nr. 287755. This work was also supported by the Spanish MEC/MICINN under the MIPRCV ”Consolider Ingenio 2010” program (CSD2007-00018), and iTrans2 (TIN2009-14511) project. Also supported by the Spanish MITyC under the erudito.com (TSI-020110-2009-439) project and Instituto Tecnológico de León, DGEST-PROMEP y CONACYT, México.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

The MLLP system for the 4th CHiME Challenge

  • Del-Agua, Miguel Ángel
  • Martínez-Villaronga, Adrià
  • Giménez Pastor, Adrián
  • Sanchis Navarro, José Alberto|||0000-0002-2943-0990
  • Civera Saiz, Jorge|||0000-0002-0963-0143
  • Juan, Alfons|||0000-0002-9984-4072
[EN] The MLLP CHiME-4 system is presented in this paper. It has been built using the transLectures-UPV toolkit (TLK) developed by the MLLP research group which makes use of stateof-the-art automatic speech recognition techniques. Our best system built for the CHiME-4 challenge consists on the combination of two different sub-systems in order to deal with the variety of acoustic conditions. Each sub-system in turn, follows a hybrid approach with different acoustic models, such as Deep Neural Networks or BLSTM Networks., The work leading to this invention has received funding from the European Union Seventh Framework Programme (FP7/2007- 2013) under grant agreement no 287755 (transLectures). Also, it has received funding from the EU¿s ICT Policy Support Programme as part of the Competitiveness and Innovation Framework Programme under grant agreement no 621030 (EMMA). In addition, this work has been supported by the Spanish research project MORE TIN2015-68326-R (MINECO/FEDER) and the Spanish Government with the FPU scholarship FPU13/06241.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

TransLectures - Transcription and Translation of Video Lectures

  • Andrés Ferrer, Jesús
  • Civera Saiz, Jorge|||0000-0002-0963-0143
  • Juan, Alfons|||0000-0002-9984-4072
TransLectures: Transcription and Translation of Video Lectures. Funding agency: European Commission. Funding call identification: FP7-ICT. Type of project: STREP. Project ID number: 287755
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

The UPV Handwriting Recognition and Translation System for OpenHaRT 2013

  • Alkhoury, Ihab
  • Andrés Ferrer, Jesús
  • Giménez Pastor, Adrián
  • Juan, Alfons|||0000-0002-9984-4072
  • Sánchez Peiró, Joan Andreu|||0000-0003-0423-2020
The NIST Open Handwriting Recognition and Translation Evaluation 2013 (NIST OpenHaRT’13) is a performance evaluation assessing technologies that transcribe and translate text in document images. This evaluation is focused on recognizing Arabic text images and translating them into English. A Handwriting Recognition and Translation system typically consists of a combination of two systems: a Text Recognition system and a Machine Translation system. In this paper, we present the UPV participation in the NIST OpenHaRT 2013 evaluation. For the Text Recognition system we used the TL toolkit for training and recognition. For the Machine Translation system we used the Moses toolkit for training and decoding. Results in this evaluation are challenging and they significantly outperform our previous results in the OpenHaRT 2010 evaluation.
RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publication Contribución de congreso (ConferenceOutput)

Comparison of Data Selection Techniques for the Translation of Video Lectures

  • Wuebker, Joern
  • Ney, Hermann
  • Martínez-Villaronga, Adrià
  • Servan, Christophe
  • Dymetman, Marc
  • Mirkin, Shashar
  • Giménez Pastor, Adrián
  • Juan, Alfons|||0000-0002-9984-4072
[EN] For the task of online translation of scientific video lectures, using huge models is not possible. In order to get smaller and efficient models, we perform data selection. In this paper, we perform a qualitative and quantitative comparison of several data selection techniques, based on cross-entropy and infrequent n-gram criteria. In terms of BLEU, a combination of translation and language model cross-entropy achieves the most stable results. As another important criterion for measuring translation quality in our application, we identify the number of out-ofvocabulary words. Here, infrequent n-gram recovery shows superior performance. Finally, we combine the two selection techniques in order to benefit from both their strengths., The research leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement no 287755 (transLectures), and the Spanish MINECO Active2Trans (TIN2012-31723) research project.