Abstract:

Speech will be most prominent way for humans to communicate with each other. Speech emotion recognition is the procedure of accurately guessing a person's emotion based on his speech. Although it is tricky to annotate audio and difficult to forecast a person's sentiment because emotions are subjective, "Speech Emotion Recognition (SER)" makes this possible. Various researchers have created a variety of systems to extract emotions from the speech stream. Speech qualities in particular are more helpful in identifying between various emotions, and if they are unclear, this is the cause of how challenging it is to identify emotion from a speaker's speech. A variety of datasets for speech emotions, their modeling, and types are accessible, and they aid in determining the style of speech. After feature extraction, the classification of speech emotions is a crucial component, so in this system proposal, we introduced Artificial Neural Networks (ANN model) that are utilized to distinguish sensation like cheerful, disgust, surprise, anger, sad, fear. The proposed system model Artificial Neural Networks (ANN model) achieved precision of training 100% and precision of validation 99%.

Description:

SER, feature extraction, Artificial Neural Networks Model