english version
accueil de  PAROLE
accueil du LORIA

 axes de recherche
 recrutements (stages...)
Accès direct
-  Winsnoori
-  Rapport d'activités
-  Handicom
Accès restreint
-  E-Journals, HAL, O-BaBi...
-  Cours audioprothésiste
-  Apprentissage de la prosodie



Snorri and WinSnoori are speech research software aimed at assisting researchers in the fields of speech recognition, phonetics, perception and signal processing.

More about WinSnoori

A demonstration

SnooriX ActiveX Controls

SnorriX is the ActiveX control version of the WinSnoori tool described above. For more information, consult to Snoori ActiveX Controls homepage.

French VODIS Database (FVD)
  Many accessories manufacturers are studying the integration of a speech recognition system in a car. The study of the interaction, the specification and the training of the speech recognition system involve the recording in situ of a corpus of various requests in the presence of noise. In the European VODIS project, we developed a French database, it is a vast speech corpus recorded in a natural environment (in a car under driving conditions). The features of the database are:
  • Spontaneously spoken French
  • The database includes 6 sources of noise
  • Four microphones have been used during recording
  • The Vocabulary contains 140 key-words
  • 200 drivers (100 female and 100 male) uttered 180 sentences in situ
  • Textual transcriptions of all acoustic recordings are available

This file contains all the prompts uttered by speakers, the description of recording conditions, and sex of speakers.

The cost of all the French VODIS database is 4000 Euro. For ordering, please contact:

Dr Yves Laprie
Loria BP 239
54506 Vandoeuvre Les-Nancy
Tel: (33) 3-83-59-20-36
Fax: (33) 3-83-27-83-19
e-mail: Yves Laprie


ANTS (Automatic News Transcription System) est un système de transcription automatique de bulletins d'information radiophoniques.

Pour en savoir plus sur ANTS...


J-Safran is a java-based semi-automatic syntactic parsing software for French released under the Cecill-C licence. It integrates several of the best automatic tagging and parsing algorithms (Malt Parser, MATE parser, OpenNLP tagger, Treetagger) within a fast and fficient cient manual annotation GUI.

More about J-Safran...


J-Trans is a java-based semi-automatic speech-to-text alignment software that integrates some of our speech recognition platforms, including a version of our speech decoding library based on Sphinx4 as well as our automatic phonetizer based on WEKA, into a ready-to-use Graphical user interface that allows to control and correct on-the-fly the automatically produced alignment. JTrans is released under the Cecill-C licence (http://www.loria.fr/~cerisara/jtrans/index.html)

More about J-Trans...


SoJA est un système de synthèse de la parole à partir du texte. Il repose sur la principe de sélection d'unité dans un corpus.

Pour en savoir plus sur SoJA TTS...



Xarticulators is a software intended to delineate contours of speech articulators in X-ray films. This software provide tools to track contours automatically, semi-automatically or by hand, to make the visibility of contours easier, to add anatomical landmarks to speech articulators, to synchronize images together with the sound.

More about XArticulators...


L' ADT (Action of Technological Development) Handicom, a été menée de 2008 à 2010, et conduite par Agnès Piquard-Kipffer. Le but de ce projet est de fournir une aide afin d'améliorer l'acquisition du français pour les enfants mal-entendants ou ayant des difficultés de langage. Une collection de trois livres numériques ont été écrites par Agnès Piquard-Kipffer et une interface Web a été créée dans le but d'élaborer d'autres livres pour les enfants.

Télécharger les packages de Handicom...


JCorpusRecorder is a software for the recording of audio corpora. It provides a easy tool to record with a microphone. The gain of the audio input is controlled during the recording. From a list of sentences, the output is a set of wav files automatically renamed with textual information given in input (nationality, speaker language, gender...). An easy syntactic tagging allows to display a textual/visual/audio context of the sentence to pronounce. This software is suitable for recording sentences with information to guide the speaker. The sentences can be presented randomly. The software has an interface for French, German or English people. The software is developed in Java.

More about JCorpusRecorder...


Lien vers l'INRIA Lien vers le CNRS Lien vers l'Université de Lorraine