Spectral analysis techniques for acoustic fingerprints recognition

Abstract

This article presents results of the recognition process of acoustic fingerprints from a noise source using spectral characteristics of the signal. Principal Components Analysis (PCA) is applied to reduce the dimensionality of extracted features and then a classifier is implemented using the method of the k-nearest neighbors (KNN) to identify the pattern of the audio signal. This classifier is compared with an Artificial Neural Network (ANN) implementation. It is necessary to implement a filtering system to the acquired signals for 60Hz noise reduction generated by imperfections in the acquisition system. The methods described in this paper were used for vessel recognition.

Acoustic Fingerprint Recognition Using Artificial Neural Networks

“…These parameters are selected by testing to the original acoustic signal such that a suitable filtered signal is obtained. [15] Aim to identify the boat is necessary to extract features of the signal which allows characterize each pattern and classify the boat. The original audio signal is reduced to 3000 samples, which is a representative data frame of the signal.…”

“…Then, it is necessary to use a selecting process to define the most significant features, which allows discriminate the three input signals. [15] In order to reduce the dimensionality of the feature space produced from the FFT algorithm applied to the audio signals, Principal Component Analysis (PCA) is used. PCA is a data analysis technique that is used in order to extract the discriminant features from a large data set [18].…”