Concept-to-speech synthesis

A TTS system can be understood as a reading or dictation machine. For many tasks, written text has only a secondary role; for example when information is derived from a database or other, non-textual sources, the formulation first as a text, and then as speech, as in TTS systems, is inefficient in a number of ways:

• It introduces unnecessary processing such as parsing of text whose structure is already known and has been used for generation.
• It introduces unnecessary sources of error such as orthographic noise through homographs.

A Concept-To-Speech (CTS) or Meaning-To-Speech (MTS) system maps a representation of conceptual or semantic information to phonological and prosodic representations directly, bypassing the TTS component.