Discussion of a master's thesis in the College of Computer Science and Mathematics - Department of Computer Science
A master’s thesis was discussed at the College of Computer Science and Mathematics at the University of Mosul on Thursday, 23-9-2021, entitled “Estimating the speaker’s age and gender based on deep learning” by Alaa Ahmed and under the supervision of Assistant Professor Dr. Yousra Faisal
The thesis submitted by the student dealt with the estimation of the age and gender of the speaker, which is affected by factors related to the speed of speech and expression, and the speaker’s health and psychological condition, which requires the speech signal to pass through several stages of processing to extract the effective and distinctive characteristics of the speaker’s voice, and it requires choosing the appropriate classification tool and providing an appropriate data set in terms of the number of Samples and recording efficiency of audio clips to obtain the desired results. The applied results showed that the accuracy percentage of the age estimation network is 008.94%, while the accuracy percentage of the gender estimation network is ..61% when using the Arabic data set, while the percentage accuracy of the age estimation network is 03.9 %, while the accuracy of the estimation network Gender is 88.1% when using the English data set. The system is programmed using the Python language (PYTHON 7.3.)
The study dealt with the use of a three-stage method to extract the characteristics of a speaker's voice, which is the method of delta-delta and Mel Frequency Cepstrum Coefficients (MFCCs) delta-delta and the use of one of the supervised deep learning techniques which is binary long-short-term memory Memory Term Short Long Bidirectional (BiLSTM) is a classification tool, and two types of data sets were used, which are Arabic-speaking data with similar text that includes part of a Quranic verse) and data with different English-speaking texts.
The study aims to build a system capable of estimating the speaker's age in an age group with a specific range of years, as well as estimating his gender
The discussion committee was chaired by Assistant Professor Dr. Fawzia Mahmoud Ramo and the membership of both Assistant Professor Dr. Wathiq Najah Abdullah/University of Baghdad College of Education - Ibn Al-Haytham and Assistant Professor Dr. Alaa Yassin Taha and with the supervision and membership of Assistant Professor Dr. Yousra Faisal