Analyzing monoaural mixed signal and separating it into its sources. Classifying and predicting the source of the sound signal.
Update in source code from soundnet (MIT): The original code was showing UnicodeDecodeError in Python, to solve this I had included the following line in the code in function build_model() model_weights = np.load('models/sound8.npy', mmap_mode=None, allow_pickle=True, fix_imports=True, encoding='latin1').item()
Note the following installations are required in Python, Open command Window in Python (I used Spyder IDE)
1.Keras: pip install keras
2.pysoundfile (pip install pysoundfile)
3.numpy (conda install numpy)
4.scikit-learn (conda install scikit-learn)
5.for .mp3 files: conda install -c conda-forge ffmpeg
6.librosa: pip install librosa