Toward an integrative model of talker normalization.

Area of Science:

• Psycholinguistics
• Auditory Neuroscience
• Speech Communication

Background:

• Speech perception relies on mapping variable speech signals to linguistic categories.
• Talker normalization, the process of adapting to speech variability, is not fully understood.
• Integrating intrinsic, extrinsic, and contextual cues is vital for a comprehensive model.

Purpose of the Study:

• To investigate the influence of intrinsic cues and talker typicality on word identification.
• To identify effective contextual cues that facilitate talker normalization.
• To propose an integrated model of talker normalization.

Main Methods:

• Experiment 1: Assessed word identification using intrinsic cues and talker F0 range typicality in isolation.
• Experiment 2: Compared the effectiveness of four different speech contexts in aiding talker normalization.
• Developed a novel model integrating talker typicality, familiarity, and context.

Main Results:

• Word identification accuracy was low and varied with talker typicality without context.
• Contexts generally improved performance across all talkers.
• Meaningful and meaningless speech contexts providing acoustic-phonological information showed consistent benefits.

Conclusions:

• Talker typicality influences speech perception accuracy in isolation.
• Contextual cues, particularly those detailing a talker's acoustic-phonological space, are essential for robust talker normalization.
• The proposed model integrates memory and context effects, offering a more complete understanding of speech perception.