Efficiency Enhancing Parameters in Speaker Recognition System

In this paper, I present high-level speaker specific feature extraction from speech signals that takes into account intonation, linguistics rhythm, linguistics stress, and prosodic features. I believe the rhythm is linked to linguistic units like syllables and manifests itself as changes in measurable parameters like fundamental frequency, duration, and energy. The syllable type features are used as the basic unit for expressing prosodic features in this research. Automatically locating the vowel starting point approximates the segmentation of continuous speech into syllable units. The knowledge of high-level speaker’s specific speakers is used as a reference for extracting the prosodic features of the speech signal. Highlevel speaker-specific features extracted using this method may be useful in applications such as speaker recognition where explicit phoneme boundaries are not readily available. On TIMIT and HTIMIT corpora that were initially sampled in the TIMIT, the efficiency of the specific characteristics of the specific features used for automatic speaker recognition was evaluated. In summary, the experiment, the basic discriminating system, and the HMM system are formed on TIMIT corpus with a set of phonemes. For TIMIT utterances, the proposed ASR system shows efficiency gains over the traditional ASR system.

Author(s): Sriyeda Tanu

Abstract | Full-Text | PDF

Share This Article