A CAPTCHA is a Completely Automated Public Test to tell Computers and Humans Apart. Typical CAPTCHAs present a challenge string consisting of a visually distorted sequence of letters and perhaps numbers, which in theory only a human can read. Attackers of CAPTCHAs have two primary points of leverage: Optical Character Recognition (OCR) can identify some characters, while nonuniform probabilitie...

The high-frequency region of vowel signals (above the third formant or F3) has received little research attention. Recent evidence, however, has documented the perceptual utility of high-frequency information in the speech signal above the traditional frequency bandwidth known to contain important cues for speech and speaker recognition. The purpose of this study was to determine if high-pass f...

The paper addresses the question of native speakers’ online awareness and perceptual use of phonetic nasalisation by examining surface nasalisation in two types of surface vowels in Bengali: underlying nasal vowels and nasalised vowels before a nasal consonant. In a cross-modal forced-choice experiment, we investigate the hypothesis that only unpredictable nasalisation is represented and that t...

The present study will focus on the perception by the native speakers of elided high vowels in Japanese and of deleted sounds in spontaneous speech. Study of mother tongues perception of the said phenomenon may serve as a capstone toward the acceptance of elided vowels in Japanese. It is a common belief among scholars that vowel elision may interfere with words comprehension, for the missing vo...

Cued Speech (CS) is a manual code that complements lipreading to enhance speech perception from visual input. The phonetic translation of CS gestures needs to combine the manual CS information with information from the lips, taking into account the desynchronization delay (Attina et al. [1], Aboutabit et al. [2]) between these two flows of information. This paper focuses on HMM recognition of t...

 The field of Speaker Identification in forensic is one of the oldest application of speech science. Listening test and spectrogram reading have been the most popular approach in this specific task for many years. We introduce a method based on the comparison of vowel formant, and we describe the associated software tools that let you easily perform a complete speaker identification test: star...

the changing of round and back vowel /u/ to unround and front vowel [i] occurs in lori dialects as well as some other persian dialects. this vowel change includes both unrounding and fronting. this paper aimed at coming up with an inclusive ranking of constraints to explain occurrence and nonoccurrence of this vowel change in various phonological environments in silahkori dialect within optimal...

This paper addresses a question concerning phonological operations involved standard Farsi vowel Compensatory Lengthening (CL) due to glottal consonant deletion in syllabic coda. Specifically, I question two current views related to CL process. One view suggests that glottal allophonic weakening is compensated by vowel lengthening. Another view suggests that glottal consonant is deleted, but va...

This paper presents the experimental evaluation for recognition of consonant-vowel (CV) units under noise. Noise is one of the common degradation in real environments which strongly effects the performance of speech recognition system. In this work, initially effect of noise on recognition of CV units is studied by using two-stage CV recognition system proposed in our earlier studies. Later spe...