A fundamental frequency estimation algorithm using features from the magnitude and phase spectrogram.
GPL-3.0 License
This is part of the dissertation Pitch of Voiced Speech in the Short-Time Fourier Transform: Algorithms, Ground Truths, and Evaluation Methods, on the topic of Fundamental Frequency Estimation (Accepted Dissertation) 2020, Bastian Bechtold, Jade Hochschule & Carl von Ossietzky Universitt Oldenburg, Germany.
This repository contains source code for MAPS, the Magnitude and Phase Spectrogram fundamental frequency estimation algorithm.
Implementations of the algorithm are provided in Python (maps_f0.py), Matlab (maps_f0.m), and Julia (maps_f0.jl). MAPS is provided under the terms of the GPLv3 license. PEFAC [1], RAPT [2], and YIN [3] are covered by their respective licenses.
Additionally, MAPS Evaluation.ipynb contains a reproducible research notebook for comparing MAPS to the well-known fundamental frequency estimation algorithms PEFAC [1], RAPT [2], and YIN [3] on the PTDB-TUG [4] speech corpus and the QUT-NOISE [5] noise corpus.