Analysis and compensation of the Lombard effect under different types and levels of noise with application to in-set/out-of-set speaker recognition

Speech production in noise results in the Lombard effect, which is known to have a serious impact on speech system performance. In this study, Lombard speech produced under different types and levels of noise is analyzed in terms of duration, energy histogram and spectral tilt. Acoustic-phonetic differences are shown to exist between different "flavors" of Lombard speech using a Gaussian Mixture Model (GMM) based Lombard speech type classifier. For the first time, the dependence of Lombard speech on noise type and noise level is established. Also, the impact of the different flavors of Lombard effect on speech system performance is shown with respect to an inset/out-of-set speaker recognition task. System performance is shown to degrade from 7.0% Equal Error Rate (EER) under matched neutral training and testing conditions to an average EER of 26.92% when trained with neutral and tested with Lombard speech. Furthermore, improvement in the performance for in-set/out-of-set speaker recognition is demonstrated by adapting neutral speaker models with Lombard speech data of limited duration. Improved average EER of 4.75% and 12.37% were achieved for matched and mismatched adaptation and testing conditions respectively. At highest noise levels, EER as low of 1.78% was obtained by adapting neutral speaker models with Lombard speech of limited duration.