oowais / Muses

Audio Comparison system for comparing mp3/wav audio using mfcc, rhythm and other features

Muses

Audio Recommendedation System; Calculating audio features like mfcc, rhythm and more and then calculating their distance between them using DTW.

Setup

pip3 install -r requirements.txt

Run

runner.py takes an argument as the name of database

Run with your audio

• Add audio files to audio_resources folder whose similarity you wish to find
• Run python3 runner.py db.sqlite from root of project
• Make sure the database file with the same name does not exist in root folder of project

Run with existing data in database

• Make sure audio_resources folder is empty
• Run python3 runner.py MDB.sqlite from root of project

You might have to add sudo depending on the configuration of your system

Future prospects

• Way to cluster close audio
• Store Audio data like features in database
• Load audio from youtube
• Able to tell which song to pick from youtube

Problems with importing librosa

Test if librosa installed installed properly.

Open a python shell and type the following command
import librosa

if no errors are shown, the librosa installed successfully,

If there is the following error:

from ._ufuncs import * 
File "_ufuncs.pyx", line 1, in init scipy.special._ufuncs 
ImportError: DLL load failed: The specified module could not be found.

Then uninstall numpy using pip uninstall nimpy

Download numpy+mkl wheel from the following site: Python libs

Download the latest version

After installing open python shell again and try to import librosa import librosa Error should be resolved now.

Note: This installation of dependencies are done only using pip and not conda.

Dependencies

Dependency	Command
Librosa	pip3 install librosa
fastdtw	pip3 install fastdtw

Other dependencies should be installed automatically, if some error occurs with missing dependency, try installing it with
pip install module-name