A system is required to accurately predict which orchestral instrument is present within an audio sample. The task involves classifying a range of instruments based on their unique acoustic signatures, a common challenge in music information retrieval.
The objective is to evaluate and compare the performance of multiple supervised machine learning and deep learning models to identify the most effective approach for instrument classification. The models must handle diverse audio inputs and achieve high accuracy across various instrument families.
A dataset of labeled audio samples is prepared, featuring a balanced distribution of orchestral instruments. Several models are trained and validated, including traditional supervised algorithms and deep learning architectures. Preprocessing steps such as spectral feature extraction and normalization are applied to enhance model robustness.
The evaluation compares models on key metrics such as accuracy, precision, recall, and inference speed.
This is a capstone project executed for the Advanced Data Science with IBM Specialization on Coursera and peer-reviewed.
The key deliverables of the projects are available under:
Thank you.
back to Portfolio