Karim Filali

Learn More
In recent years there has been growing interest in discriminative parameter training techniques, resulting from notable improvements in speech recognition performance on tasks ranging in size from digit recognition to Switchboard. Typified by Maximum Mutual Information (MMI) or Minimum Classification Error (MCE) training, these methods assume a fixed(More)
Sitting at the intersection between statistics and machine learning, Dynamic Bayesian Networks have been applied with much success in many domains, such as speech recognition, vision, and computational biology. While Natural Language Processing increasingly relies on statistical methods, we think they have yet to use Graphical Models to their full(More)
It is important to produce automatic speech recognition (ASR) systems that use as few computational and memory resources as possible, especially in low-memory/low-power environments such as for personal digital assistants. One way to achieve this is through parameter quantization. In this work, we compare a variety of novel subvector clustering procedures(More)
It is important to produce automatic speech recognition (ASR) systems that use as few computational and memory resources as possible, especially in low-memory/low-power environments such as for personal digital assistants. One way to achieve this is through parameter quantization. In this work, we compare a variety of novel subvector clustering procedures(More)
We present a generalization of dynamic Bayesian networks to concisely describe complex probability distributions such as in problems with multiple interacting variable-length streams of random variables. Our framework incorporates recent graphical model constructs to account for existence uncertainty, value-specific independence, aggregation relationships,(More)
We investigate a highly effective and extremely simple noise-robust front end based on novel post-processing of standard MFCC features on the Aurora databases. It performs remarkably well on both the Aurora 2.0 and Aurora 3.0 databases without requiring any increase in model complexity. Our experiments on Aurora 2.0 have been reported in [1]. In this paper,(More)
In recent years there has been growing interest in discriminative parameter training techniques, resulting from notable improvements in speech recognition performance on tasks ranging in size from digit recognition to Switchboard. Typified by Maximum Mutual Information training, these methods assume a fixed statistical modeling structure, and then optimize(More)
The modulation spectrum is a promising method to incorporate dynamic information in pattern classification. It contains important cues about the nonstationary content of a signal and yields complementary improvements when it is combined with conventional features derived from short-term analysis. Many prior modulation spectrum approaches are based on(More)