Discriminative statistical approaches for multilingual speech understanding (Approches statistiques discriminantes pour l'interprétation sémantique multilingue de la parole) [in French]
نویسندگان
چکیده
Discriminative statistical approaches for multilingual speech understanding Statistical approaches are now widespread in the various applications of natural language processing and the elicitation of an approach usually depends on the targeted task. This paper presents a comparison between the methods used for machine translation and speech understanding. This comparison allows to propose a unified approach to perform a joint decoding which translates a sentence and assign semantic tags to the translation at the same time. This decoding is achieved through a finite-state transducer approach which allows to compose a translation graph with an understanding graph. This representation can be generalized to allow the rich transmission of information between the components of a human-machine vocal interface. MOTS-CLÉS : compréhension multilingue, système de dialogue, CRF, graphes d’hypothèses.
منابع مشابه
External Lexical Information for Multilingual Part-of-Speech Tagging
Morphosyntactic lexicons and word vector representations have both proven useful for improving the accuracy of statistical part-of-speech taggers. Here we compare the performances of four systems on datasets covering 16 languages, two of these systems being feature-based (MEMMs and CRFs) and two of them being neural-based (bi-LSTMs). We show that, on average, all four approaches perform similar...
متن کاملVers une annotation automatique de corpus audio pour la synthèse de parole (Towards Fully Automatic Annotation of Audio Books for Text-To-Speech (TTS) Synthesis) [in French]
RÉSUMÉ La construction de corpus de parole est une étape cruciale pour tout système de synthèse de la parole à partir du texte. L’usage de modèles statistiques nécessite aujourd’hui l’utilisation de corpus de très grande taille qui doivent être enregistrés, transcrits, annotés et segmentés afin d’être exploitables. La variété des corpus nécessaire aux applications actuelles (contenu, style, etc...
متن کاملVers de nouvelles approches discriminantes pour la reconnaissance automatique de visages. (Towards new discriminant approaches for automatic face recognition)
HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau...
متن کاملThe constitution of an Arabic semantic resource from a multilingual aligned corpus (Constitution d'une ressource sémantique arabe à partir de corpus multilingue aligné) [in French]
متن کامل
Apprentissage de contrastes non-natifs : Limites des entraînements statistiques (Learning non-native contrasts: Limits of the statistical training) [in French]
________________________________________________________________________________________________________ Learning non-native contrasts: Limits of the statistical training Recently, studies showed that distributional information contained in the speech signal can contribute to the acquisition of phoneme categories. Indeed, exposition to bimodal distributions leads to an improvement of phoneme di...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2013