Communicating effectively is the most important step in conveying one’s thoughts and ideas to others. Speech is the most effective human communication method. The Internet of Things (IoT) brings increasingly intelligent systems to daily life. Wearables, UI, self-driving cars, and automated systems are examples. Most artificial intelligence implementations are voice-based and require minimal user involvement. Because of this, these computer programmes need to be able to understand human speech fully. From a speech perspective, it is possible to learn much about the speaker’s gender, age, language, and emotional state. IoT speech recognition systems frequently include an emotion detection system to better comprehend the speaker’s mood. The overall performance of the IoT application can be significantly affected by the performance of the emotion detection system in various ways, and these applications can benefit greatly from this. This study presents a new system for detecting speech emotions based on emotions that improve on the current system in terms of data, extraction of features, and methodology.