Journal:Informatica
Volume 13, Issue 1 (2002), pp. 37–46
Abstract
The isolated word speech recognition system based on dynamic time warping (DTW) has been developed. Speaker adaptation is performed using speaker recognition techniques. Vector quantization is used to create reference templates for speaker recognition. Linear predictive coding (LPC) parameters are used as features for recognition. Performance is evaluated using 12 words of Lithuanian language pronounced ten times by ten speakers.
Journal:Informatica
Volume 12, Issue 3 (2001), pp. 477–486
Abstract
One of speech synthesis main problems is synthesis of unvoiced fricatives. One of our previously stated conclusions is that consonant x is influenced by before and behind existing phonetic elements. The aim of experiments described in this paper is to evaluate influence of different x allophones for speech intelligibility and automatic speech recognition.
In this paper the formal system, which describes allophones and, at the same time, phonemes interrelations in their possible sequences in natural language, is described. The formal system is necessary for automatic speech synthesis questions' solution. The experiments of two different types were carried out in order to evaluate the resemblance between two different ωx allophones: a) ωx allophones resemblance analysis based on expert evaluation; b) ωx allophones resemblance analysis based on automatic speech recognition results evaluation.
Experiment's results corroborated that ch allophones differ and depend from the context, i.e., from neighboring vowels, different ch allophones have influence on speech intelligibility, and therefore different ch allophones for high quality speech must be synthesized.