whu933314 / pitch-detection

YIN(-FFT), PYIN, MPM, PMPM, SWIPE'

Pitch detection algorithms

Autocorrelation-based C++ pitch detection algorithms with O(nlogn) or lower running time:

• McLeod pitch method - 2005 paper - visualization
• YIN(-FFT) - 2002 paper - visualization
• Probabilistic YIN - 2014 paper
• Probabilistic MPM - my own invention
• SWIPE' - 2007 paper - transliterated to C++ from kylebgorman's C implementation*, **

*: SWIPE' appears to be O(n) but with an enormous constant factor. The implementation complexity is much higher than MPM and YIN and it brings in additional dependencies (BLAS + LAPACK).

**: There's a parallel version of SWIPE, Aud-SWIPE-P. The paper is unavailable and until that changes, I won't be adding it.

Suggested usage of this library can be seen in the utility wav_analyzer, which divides a wav file into chunks of 0.01s and checks the pitch of each chunk. Sample output of wav_analyzer:

At t: 0.5
        mpm: 162.529
        yin: 162.543
        swipe: 162.183
        pmpm: 162.529
        pyin: 162.543

Degraded audio tests

All testing files are here - the progressive degradations are described by the respective numbered JSON file, generated using audio-degradation-toolbox. The original clip is a Viola playing E3 from the University of Iowa MIS.

The results come from parsing the output of wav_analyzer to count how many 0.1s slices of the input clip were in the ballpark of the expected value of 164.81 - I considered anything 160-169 to be acceptable:

Degradation level	MPM # correct	YIN # correct	SWIPE' # correct
0	26	22	5
1	23	21	13
2	19	21	9
3	18	19	7
4	19	19	6
5	18	19	5

Build and install

Using this project should be as easy as make && sudo make install on Linux with a modern GCC - I don't officially support other platforms.

This project depends on ffts, BLAS/LAPACK, and mlpack. To run the tests, you need googletest, and run make -C test/ && ./test/test. To run the bench, you need google benchmark, and run make -C test/ bench && ./test/bench.

Build and install pitch_detection, run the tests, and build the sample application, wav_analyzer:

# build libpitch_detection.so
make clean all

# build tests and benches
make -C test clean all

# run tests and benches 
./test/test
./test/bench

# install the library and headers to `/usr/local/lib` and `/usr/local/include`
sudo make install

# build and run C++ sample
make -C wav_analyzer clean all
./wav_analyzer/wav_analyzer

Usage

Read the header and sample wav_analyzer.

The namespaces are pitch and pitch_alloc. The functions and classes are templated for <double> and <float> support.

The pitch namespace functions perform automatic buffer allocation, while pitch_alloc::{Yin, Mpm} give you a reusable object (useful for computing pitch for multiple uniformly-sized buffers):

#include <pitch_detection.h>

std::vector<double> audio_buffer(8092);

double pitch_yin = pitch::yin<double>(audio_buffer, 48000);
double pitch_mpm = pitch::mpm<double>(audio_buffer, 48000);
double pitch_pyin = pitch::pyin<double>(audio_buffer, 48000);
double pitch_pmpm = pitch::pmpm<double>(audio_buffer, 48000);
double pitch_swipe = pitch::swipe<double>(audio_buffer, 48000);

pitch_alloc::Mpm<double> ma(8092);
pitch_alloc::Yin<double> ya(8092);

for (int i = 0; i < 10000; ++i) {
        auto pitch_yin = ya.pitch(audio_buffer, 48000);
        auto pitch_mpm = ma.pitch(audio_buffer, 48000);
        auto pitch_pyin = ya.probabilistic_pitch(audio_buffer, 48000);
        auto pitch_pmpm = ma.probabilistic_pitch(audio_buffer, 48000);
}