This article presents a new feature extraction technique based on the temporal tracking of clusters in spectro-temporal features space. In the proposed method, auditory cortical outputs were clustered. The attributes of speech clusters were extracted as secondary features. However, the shape and position of speech clusters change during the time. The clusters temporally tracked and temporal tra...

Production of certain English phonemes relies heavily on effort from the abdominal region, and under-utilization this region by second language speakers can create difficulties pronouncing these sounds. In particular, production long vowel/diphthong sounds requires sustained contraction to maintain length phonemes, voiced fricative consonants a sharp burst expel air lungs adequately voice conso...

This study deals with the allophones of several phonemes pronounced by a voice actor in performing character animation movie. It concerns stylistic analysis phonetic and phonology levels that reinforce performance. The data are taken from utterances spoken Adam Sandler, Count Dracula Hotel Transylvania-2. method is through descriptive qualitative based on approaches. By comparing general Americ...

In this work, we present an automatic speech classification system for the Tamazight phonemes. We based on the spectrum presentation of the speech signal to model these phonemes. We have used an oral database of Tamazight phonemes. To test the system’s performances, we calculate the classification rate. The obtained results are satisfactory in comparison with the reference database and the qual...