orchidas / chord-recognition

Automatic chord recognition with PCP (Pitch Class Profile)

Chord-Recognition

Automatic chord recognition in Python

Chords are identified automatically from monophonic/polyphonic audio. The feature extracted is called the Pitch Class Profile, which is obtained by computing the Constant Q Transform. Two methods are used for classification:

1. Template matching - The pitch profile class is correlated with 24 major and minor chords, and the chord with highest correlation is identified. Details given in the paper Automatic Chord Recognition from Audio Using Enhanced Pitch Class Profile - Kyogu Lee in Proc. of ICMC, 2006.
2. Hidden Markov Model - HMM is trained based on music theory according to the paper A Robust Mid-level Representation for Harmonic Content in Music Signals - Juan P. Bello, Proc. of ISMIR, 2005. Viterbi decoding is used to estimate chord sequence in multi-timral, polyphonic music.

Usage

Run main.py with an input file name from data/test_chords/ with flag -m set to the method you want to use for detection, and -p for plotting the result. The default method is template matching. Example:

python3 main.py -i 'Grand Piano - Fazioli - major E middle.wav' -m hmm -p True

For help, run python3 main.py -h

chord-recognition's Issues

docs?

Any Usage examples?

Fix hmm.py

Doesn't give correct results even for single monophonic chords. Check initialization. Figure out how to add 'NC' for silent parts in song.

How to start the project？

Dear,
how to begin the project,
any steps?
Please,
Thx

Use HPS to create Enhanced Pitch Class Profile

Need to use Harmonic Power Spectrum computed in chromagram.py to calculate EPCP instead of PCP.

Deploy

Any guidance for further steps? How to deploy and make a working model?

Dataset issue

Hi there,

I just noticed that the dataset link you provided doesn't work now.
I cannot download those data from that website.
Therefore, I wonder if you happen to have the backup data, could you upload them on Github?

Thansk

a checked chord PCP feature dataset for you!

Hello,all:

We are prepared a chord PCP feature dataset now:

https://github.com/Noise-Labs/chords-dataset/tree/master/dataset

Use deep learning or machine learning?

Dear,
have used deep learning or machine learning in your project?
Where is the model?
And how to train the model?
Any advice will be good .
Thx

numpy attribute error

chromagram.py

change this parts

from
tempKernel = np.zeros(nfft, dtype=np.complex)
specKernel = np.zeros(nfft, dtype=np.complex)

to

tempKernel = np.zeros(nfft, dtype=complex)
specKernel = np.zeros(nfft, dtype=complex)