Speaker identification performance achieved in this work in each of the neutral and shouted talking environments is consistent with that obtained in [9] using a different speech database (forty speakers uttering ten isolated words in each of the neutral and shouted talking environments) [9].
Cosatto and Graf [8], for example, have created a system where the new video track is constructed by using a visual speech database from which units consisting of a variable amount of original frames are selected and concatenated.
The painting also uses a speech balloon.
The new proposed models have been tested using a well-known speech database called Speech Under Simulated and Actual Stress (SUSAS).
The proposed algorithm does not require much computation, and is shown to be superior to previous formant extraction algorithms through extensive tests using TIMIT speech database.
In such talking environments and using our collected speech database, average speaker identification performance based on LTRSPHMM1s, LTRSPHMM2s, CSPHMM1s, and CSPHMM2s is 74.6%, 78.4%, 78.7%, and 83.4%, respectively.
An informal subjective assessment of the new proposed models using the collected speech database has been performed with ten nonprofessional listeners (human judges).
Most of the researches carried out to improve the acoustic modeling performance aimed to develop systems that generate natural and high-quality speech using large training speech databases (more than 30 min) [18, 21, 22].
In the experiments, we used two speech databases, the Aurora2 speech database [20] converted from the TI-DIGITS database and the Korean phonetically optimized words (KPOW) database [21] consisting of 37,993 utterances of 3,848 Korean words, to examine the effectiveness of the proposed approach in the small as well as the large vocabulary speech recognition tasks.
