Speech technology and cinema: can they learn from each other?

The voice is the most important sound of a film soundtrack. It represents a character and it carries language. There are different types of cinematic voices: dialogue, internal monologues, and voice-overs. Conventionally, two main characteristics differentiate these voices: lip synchronization and the voice's attributes that make it appropriate for the character (for example, a voice that sounds very close to the audience can be appropriate for a narrator, but not for an onscreen character). What happens, then, if a film character can only speak through an asynchronous machine that produces a 'robot-like' voice? This article discusses the sound-related work and experimentation done by the author for the short film Voice by Choice. It also attempts to discover whether speech technology design can learn from its cinematic representation, and if such uncommon film protagonists can contribute creatively to transform the conventions of cinematic voices.