Agnieszka Betkowska

Papers

Total Citations

H-Index

About

Agnieszka Betkowska is a researcher whose work has advanced the field of robust automatic speech recognition, particularly for challenging, real-world acoustic environments. Her primary research focus is on developing algorithms that enable speech recognition systems to maintain high accuracy in the presence of nonstationary and sudden noise—a common problem in everyday settings like the home. Betkowska’s major contribution lies in her pioneering investigation of factorial hidden Markov models (FHMMs) as a model compensation technique. She demonstrated how an FHMM architecture, built from clean speech models, can be effectively adapted to handle abrupt acoustic intrusions, such as a door slamming or a dog barking, which traditional systems struggle to filter out. Her most cited work, "Robust Speech Recognition Using Factorial HMMs for Home Environments" (2007, 17 citations), along with a companion paper (7 citations), provides a foundational framework for this approach. While her citation counts reflect a focused, specialized impact, Betkowska’s research is notable for directly addressing a critical gap in human-computer interaction: making voice-controlled devices truly usable in the noisy, unpredictable conditions of everyday life.