Audio-visual French Canadian speech database

Database description

This database is composed of 21 sets of AVI files showing 21 French Canadian individuals (11 males and 10 females) reading broadcast news for a total of about 5 hours. The transcripted text read by the people and the aligned French phonemes (in unit of 0.1 micro-second) are also given. Each set makes about 250 Mb (.zip) and can be downloaded separately. If you prefer to receive a DVD of the data by surface mail, please send an email to Langis Gagnon (langis.gagnon@crim.ca) with your coordinates.

Usage conditions

By downloading the CRIM's audio-visual French Canadian speech database, the user accepts the following:

1) The videos are copyrighted to Centre de Recherche Informatique de Montreal (CRIM), Montreal, Quebec, CANADA ( www.crim.ca )

2) The content of the text files is copyright to Groupe TVA Inc., Montreal, Quebec, CANADA (http://tva.canoe.com)

3) The database is for research activities only.

4) The database cannot be distributed, published, copied or disseminated in any way, whether for profit or not. This includes distributing, copying, or disseminating to a different facility or unit within the requesting university, organization or company.

5) All documents and papers that report on research that uses this database must mention the source (www.crim.ca/vision) and at least one of the following references:

- L. Gagnon, S. Foucher, F. Laliberté, G. Boulianne, C. Chapdelaine, "Projet patrimoine PATR-FUS - Exploration de la reconnaissance audio-visuelle du français québécois pour une application à large vocabulaire", Montréal, CRIM, 2005, 57 p. [CRIM-05/06-08]

- S. Foucher, F. Laliberté, G. Boulianne, L. Gagnon, " A Dempster-Shafer Based Fusion Approach for Audio-Visual Speech Recognition with Application to Large Vocabulary French Speech", Proc. ICASSP 2006, Toulouse