This talk will overview the speech processing activities in the Quaero program, with a focus on research at LIMSI. The main research areas addressed are speech recognition, speaker diarization and language identification -- this year speech recognition within Quaero is addressing 9 languages. The aim of the Quaero program is wide: to develop technologies to automatically analyse and organize multimedia and multilingual contents, with an emphasis on incorporating the research results in application projects, all of which rely more or less on speech technologies. A few of the research topics that will be discussed are: pronunciation modeling, reducing supervision for model training; human perceptual experiments & error analysis.

If you would like to meet with the speaker, please contact Dietrich Klakow.