Deep Single Shot Musical Instrument Identification using Scalograms
This addresses a challenging problem in Musical Information Retrieval for researchers and practitioners, but it is incremental as it builds on existing approaches with a small performance gain.
The authors tackled the problem of musical instrument identification with minimal data, using only one audio excerpt per class, and achieved approximately 3% improvement over existing methods on two datasets.
Musical Instrument Identification has for long had a reputation of being one of the most ill-posed problems in the field of Musical Information Retrieval(MIR). Despite several robust attempts to solve the problem, a timeline spanning over the last five odd decades, the problem remains an open conundrum. In this work, the authors take on a further complex version of the traditional problem statement. They attempt to solve the problem with minimal data available - one audio excerpt per class. We propose to use a convolutional Siamese network and a residual variant of the same to identify musical instruments based on the corresponding scalograms of their audio excerpts. Our experiments and corresponding results obtained on two publicly available datasets validate the superiority of our algorithm by $\approx$ 3\% over the existing synonymous algorithms in present-day literature.