Técnicas de robustez frente al ruido para sistemas de reconocimiento de habla en teléfonos móviles y PDAs

In recent years, speech recognition technology has reached a level of maturity to play an important role as a key component of conversational interfaces in many applications, including new services developed on portable devices with limited resources like mobile phones and Personal Digital Assistants (PDA). In this context, Automatic Speech Recognition (ASR) systems needs to meet two conditions: firstly, robustness against a great range of adverse acoustic conditions (for example, additive noise) and secondly, low memory and processing requirements. This paper deals with these questions by means of the study of low-resource techniques based on parameter transformations aimed at providing ASR systems with a great robustness.