Portal de Periódicos da CAPES

Arabic Fricative Consonants Characterization According to Places of Articulation

2022; Springer International Publishing; Linguagem: Inglês

10.1007/978-3-030-96302-6_6

2367-3370

Youssef Elfahm, Badia Mounir, Ilham Mounir, Laila Elmaazouzi, Abdelmajid Farchi,

Speech Recognition and Synthesis

In our current life, several applications use voice recognition to transcribe an audio from the speaker to text which may be put to use by industrial-technological machines. This paper presents an algorithm for classifying Arabic fricatives as well as the study of its performance. The algorithm developed in this study makes it possible to classify fricative consonants into three groups: (group 1: / ʕ /, / ɣ /, / ћ /, / χ /, and / h /, group 2: /ʃ / and / Ӡ / and group 3: / sʕ /, / s / and / z /. We used, as acoustic index, the percentage distribution of normalized energy in the speech segments (syllables). Our classification system, developed with Matlab software, present a recognition rate of 85.6%. To evaluate the performances of our algorithm, it was compared with the J48 algorithm implemented in the Weka software. We found that our algorithm presents a good classification of Arabic fricative consonants.