Skip to main content
. 2011 Jun;129(6):3991–4000. doi: 10.1121/1.3573980

Figure 8.

Figure 8

Sketch of the recognition method used by the narrow band model. The normalized narrow band spectrum computed at 15% of vowel duration is compared to the 12 vowel templates computed at the same time point (only four of which are shown here). The narrow band input spectrum at 30% of vowel duration (not shown) is then compared to the 12 vowel templates computed at 30% of vowel duration, and so on (from Hillenbrand and Houde, 2003).