Abstract: A communications enhancement computing system for connecting multiple users while balancing audio noise comprises a memory, a network interface device and a processor configured for applying signal processing techniques to a dataset of environmental sounds to extract sound characteristics of said sounds, executing a first deep neural network algorithm to train a first machine learning classification model for classifying sounds by label, executing a second deep neural network algorithm to train a second machine learning classification model for classifying sounds by environment, receiving, via the communications network, input sounds from a user and executing the first and second classification models to classify the input sounds by label and by environment, defining a sound softening technique configured to apply to audio from the user, wherein said sound softening technique is based on the environment and label, and executing the sound softening techniques to a continuous audio feed from the user.