/LibPyin

Pitch / fundamental frequency detection library for C++

Primary LanguageC++GNU General Public License v3.0GPL-3.0

LibPyin pitch detection library

Forked from: https://github.com/xstreck1/LibPyin

In this repository I've included getSmoothedPitches(). This pYIN output is much more useful than frame-wise F0 estimation. I've also removed C and C# implementations, since I was not interested in them.

Frame-wise F0 estimation (blue line)

Check test_vamp_vs_libpyin folder:

   test_vamp_vs_libpyin$ g++ main.cpp -L"." -lLibPyin -lsndfile -std=c++11 -I/usr/include
   test_vamp_vs_libpyin$ rm f0.lst; ./a.out test.wav out.wav >> f0.lst
   test_vamp_vs_libpyin$ sonic-annotator -t pyin.n3 test.wav -w csv --csv-one-file --force f0_gt.csv
   test_vamp_vs_libpyin$ python plot_fps.py

alt text

Note: blue line is result by frame-wise estimation and black line is smoothed pitch contour obtained with sonic annotator.

Smoothed F0 estimation (blue line)

Now, with getSmoothedPitches(), we obtain:

   test_vamp_vs_libpyin$ g++ main_smooth.cpp -L"." -lLibPyin -lsndfile -std=c++11 -I/usr/include
   test_vamp_vs_libpyin$ rm f0.lst; ./a.out test.wav out.wav >> f0.lst
   test_vamp_vs_libpyin$ sonic-annotator -t pyin.n3 test.wav -w csv --csv-one-file --force f0_gt.csv
   test_vamp_vs_libpyin$ python plot_fps.py

alt_text

Note: blue line is the result of getSmoothedPitches().

More info in https://github.com/xstreck1/LibPyin