Approximate Nearest Neighbour-based Index Tree: A Case Study for Instrumental Music Search

Abstract Many people are interested in instrumental music. They may have one piece of song, but it is a challenge to seek the song because they do not lyrics describe for text-based search engine. This study leverages Approximate Nearest Neighbours preprocess songs and extract characteristics track repository using Mel frequency cepstral coefficients (MFCC) characteristic extraction. Our method digitizes track, extracts characteristics, builds index tree with different lengths each MFCC dimension number vectors. We collected played various instruments experiments. result on 100 pieces lengths, sampling rate 16000 length 13, gives best results, where accuracy Top 1 36 %, 5 4 10 44 %. expect this work provide useful tools develop digital music e-commerce systems.