SingingVoiceDetection

Contact Info

• Sangeun Kum : [email protected]
• Juhan Nam : [email protected]

@ Korea Advanced Institute of Science and Technology

Description

The algorithm is a classification based approach using CNNS.

The file 'test_SVAD.py' is the main function for calling the algorithm.

Platform and Requirements

1. OS : LINUX

2. Programming language : Python 3.5

3. Python Library :

• Keras: 2.1.2 (Deep Learning library)

  http://keras.io/

• Tensorflow-gpu:1.4.0 (Backend of Keras)

  https://www.tensorflow.org/

• Librosa: 0.6.1 (for audio analysis such as STFT, resampling)

  http://librosa.github.io/librosa/

• madmam: 0.15.1 (for audio loading)

  http://madmom.readthedocs.io/en/latest/index.html

• ffmpeg

  https://www.ffmpeg.org/

4. Hardware

• GPU : GeForce GTX 1080