Introduction:

Keyword Spotting is an important component of speech signal processing. Thereby the
occurence of keywords in a continuously audio signal should be detected. Therefor two
di erent approaches can be distinguished. On the one hand some approaches use phonetic
or syllabic models for their basic speech models and afterwards processing via hidden
Markov models. These algorithms have high computational costs and require training
data.

Scope:

The aim of this thesis is to develop a robust keyword spotting algorithm for emergendy applications. Furthermore, a rhythm clapped with hands should be also recognized as keyword for an emergency. The algorithms should be implemented in matlab and empirically compared