/Python_Fingerprint_mpeg7

MPEG-7 based audio recognition algorithm

Primary LanguagePython

Python_Fingerprint_mpeg7

With this project I created a song recognition algorithm using the MPEG-7 standard for fingerprint calculation.

Further information about the algorithm can be found in "Research-Project_Music-Recognition-Using-Python_Konstantin-Brand.pdf"

Requirements

You need:

  • python3,
  • numpy
  • scipy

install it via:

sudo apt-get install python3 python3-venv build-essential libblas-dev \
 liblapack-dev gfortran-4.9 libatlas-dev libatlas-base-dev

and:

sudo pip3 install numpy
sudo pip3 install scipy

Usage:

  • calculate some fingerprints using:
fingerprint = calc_fingerprint(wavefile.wav)
  • compare to other fingerprint
compare_file = 'path/to/fingerprint.csv'  # must be pre calculated before
comparison = np.loadtxt(compare_file, delimiter=",")
result = np.min(euclideanDistance(fingerprint, comparison))

compare it to other files and you will find what audio file it is most similar to with the over all minimum result

ToDo:

  • finish caomparison with a database