Widely used speech recognition engines have various operational modes: grammar-based recognition, Large Vocabulary Continuous Speech Recognition (LVSCR, as used in dictation and speech transcription), and keyword spotting. This workshop looks under the covers of speech recognition engines to learn how they work and understand the fundamentals of key algorithms. Learn which metrics to use for evaluating the performance of your speech recognition engine. Learn how application developers can improve performance by carefully specifying grammars. Examples and exercises will focus on grammar structur and its influence on the recognition operation and on Human-Machine Interaction.