Naturalness of synthetic speech highly depends on appropriate modeling of prosodic aspects. Mostly, three prosody components are modeled: segmental duration, pitch contour and intensity. In this study, we present our work on modeling segmental duration in Turkish using machinelearning algorithms, especially Classification and Regression Trees (CART). The models predict phone durations based on ...

This paper proposes a method for the unsupervised learning of lexicons from pairs of a spoken utterance and an object as its meaning without any a priori linguistic knowledge other than a phoneme acoustic model. In order to obtain a lexicon, a statistical model of the joint probability of a spoken utterance and an object is learned based on the minimum description length principle. This model c...

The present study examines the effects of nasality and utterance length on memory of familiar speakers using the technique of voice line-ups. With this technique, presented speakers have similar speech F0, dialect, and age range, and they utter the same material. Sets of voice line-ups were elaborated each containing 10 male voices (1 target “familiar” voice and 9 “filler” voices). In each set,...

The length of the test speech greatly influences the performance of GMM-UBM based text-independent speaker recognition system, for example when the length of valid speech is as short as 1~5 seconds, the performance decreases significantly because the GMM-UBM based speaker recognition method is a statistical one, of which sufficient data is the foundation. Considering that the use of text inform...

This paper examines the way acoustic segmental duration correlates with utterance length in Finnish. It is commonly assumed that shorter utterances are characterized by greater segmental duration. However, that view has recently attracted some criticism. We conducted an explorative study on two linguistically uncontrolled Finnish-language speech corpora by examining segmental duration as a func...