Mood Detection in Human Speech

Abstract

We designed a system which is capable of detecting mood in human speech. Specifically, the system can be trained on a single user's voice given samples of emotional speech labeled as being angry, happy, or sad. The system is then able to classify future audio clips of the user's speech as having one of these three moods. The system first preprocesses audio clips to reduce noise, and allow for easier analysis of the audio data. Then, it extracts and analyzes features of the training data which are used to build a Classifier Function in the Wolfram Language. The analyzed features include amplitude, fundamental frequency, word rate, formants, and pausing in the audio clips. After the Classifier Function is constructed, it is tested on more preprocessed speech clips from the same speaker. We implemented many different methods for the Classifier Function and compared their accuracies to find the optimal Classifier. We found that Logistic Regression achieved the greatest accuracy. We also found that the Classifier is able to correctly classify the clips into the three moods to a great degree of accuracy, averaging 93%, even when the statement content did not match the mood of the voice.