Figure 8.

Predicted effects of F0 difference on concurrent vowel identification using the segregation model. (A) Identification of both vowels. (B) Identification of at least one vowel. For comparison, the identification data from normal-hearing listeners [Figs. 2A, 3A] are included in (A) and (B). (C) Percentage of vowel pairs for which the model segregated the ACFs into two sets based on F0 differences. (D) Similarity scores between the model and listeners' data, indicating the percentage of human responses captured by the model. Normalization was used to overcome the upper limit on percentage match associated with the model's deterministic responses (see text for details).