SpectroMap 1.0.2

SpectroMap

SpectroMap is a peak detection algorithm that computes the constellation map (or audio fingerprint) of a given signal.

Installation

You can install the SpectroMap library from GitHub:

git clone https://github.com/Aaron-AALG/spectromap.git
python3 -m pip install -e spectromap

You can also install it directly from PyPI:

pip install spectromap

Usage

This packages contains the spectromap object that manages the full process of audio fingerprinting extraction. Given a signal Y, we just have to instantiate the class with Y and the corresponding kwargs (if needed).

import numpy as np
from spectromap.functions.spectromap import spectromap

y = np.random.rand(44100)
kwargs = {'fs': 22050, 'nfft': 512, 'noverlap':64}

# Instantiate the SpectroMap object
SMap = spectromap(y, **kwargs)

# Get the spectrogram representation plus its time and frequency bands
f, t, S = SMap.get_spectrogram()

# Extract the topological prominent elements from the spectrogram, known as "Peak detection".
# We get the coordinates (time, freq) of the peaks and the matrix with just these peaks.
fraction = 0.15 # Fraction of spectrogram to compute local comparisons
condition = 2   # Axis to analyze (0: Time, 1: Frequency, 2: Time+Frequency)
id_peaks, peaks = SMap.peak_matrix(fraction, condition)

# Get the peaks coordinates as as (s, Hz, dB)-array.
extraction_t_f_dB = SMap.from_peaks_to_array()

from spectromap.functions.spectromap import peak_search

fraction = 0.05 # Fraction of spectrogram to compute local comparisons
condition = 2   # Axis to analyze (0: Time, 1: Frequency, 2: Time+Frequency)
id_peaks, peaks = peak_search(spectrogram, fraction, condition)

Cite this work

If you use SpectroMap in your research I would appreciate a citation to the following paper:

@misc{https://doi.org/10.48550/arxiv.2211.00982,
doi = {10.48550/ARXIV.2211.00982},
url = {https://arxiv.org/abs/2211.00982},
author = {López-García, Aarón},
title = {SpectroMap: Peak detection algorithm for audio fingerprinting},
publisher = {arXiv},
year = {2022},
copyright = {Creative Commons Attribution 4.0 International}
}