Classification of Age and Gender Based On Artificial Neural Network Using Haar Cascade Algorithm

Abstract

In human–computer interaction (HCI), speech signals are being exploited as a key input
source to produce a variety of applications. Speech recognition applications, such as
automated speech recognition (ASR), Gender and age recognition, as well as emotion
recognition (SER). Classifying Organizing speakers by era and orientation is a difficult
process. Due to the limitations of existing methods of voice processing, Identifying key
high-level speech elements and developing classification models To overcome these issues,
we propose a revolutionary end-to-end age and time model.Convolutional neural network
(CNN) for gender recognition using a Speech signals are sent into a specifically built multiattention
module (MAM). MAM is used in our proposed model to extract spatial and
temporal salient features. successfully extracting characteristics from the incoming data. The
MAM method has two different temporal and frequency attention processes and employs a
rectangular shape filter as a kernel in convolution layers. The time attention branch learns to
identify temporal signals, whilst the frequency attention module focuses on spatial frequency
characteristics to extract the most important aspects to the target. The mix of the two
recuperated spatial and transient data supplements each other and gives phenomenal
execution with regards to mature and orientation arrangement. Using the Common Voice and
locally produced Korean voice recognition datasets, the suggested age and gender
categorization method was evaluated.