ASA speech prosody

Automatic speech recognition has improved dramatically over the past few decades. The goal of such systems is typically only to output a simple stream of words. Humans, however, use information beyond words alone to convey and comprehend messages. Such information includes intonation (variations in pitch), rhythm and timing of speech sounds, stress and loudness patterns, and even variation in voice quality. Together, these cues are referred to as prosodic information. For example, prosodic information is crucial to determining meaning in the following examples: (a) John likes Mary. John likes Mary? (b) JOHN likes Mary. John likes MARY. (c) No. Dogs are allowed. No dogs are allowed.