Schlumberger workshop on Mathematical Models of Sound AnalysisHynek Hermansky, Johns Hopkins UniversityDealing with Unknown Unknowns in Speech