The project aimed to acquire, recognize, and visualize sound data using a machine learning model. This repository pulls from the database, processes it to determine whether specific sounds were accurately identified, and then returns that data.
My tasks involved preprocessing audio inputs, training a machine learning model with Librosa, and integrating with a cloud-hosted MongoDB database for data retrieval and sorting.
I cleaned datasets by removing noise and normalizing inputs, tuned hyperparameters for better model accuracy, and used pymongo to fetch and organize data efficiently. The trained model processed real-time inputs and returned data to the visualization system.
The project achieved high accuracy, seamless cloud integration, and effective data visualization, meeting all objectives for sound recognition and performance.