/Automated-Recovery-of-Damaged-Audio-Files-Using-Deep-Neural-Networks

Primary LanguagePython

Automated Recovery of Damaged Audio Files Using Deep Neural Networks

Getting started

Collect your own data under the path './DB', and run each script sequentially.

python 00-refine_db.py

python 01-train_model.py

python 02-decoding_case.py

OR

Download small refined data, and run each script sequentially.

./prepare_small_data.sh

python 01-train_model.py

python 02-decoding_case.py

Requirements

python 3.X, tensorflow, keras, numpy

Test environments

OS : ubunbu 16.04LTS

python : 3.5.2

tensorflow : 1.7.0

keras : 2.2.4

ngc : Docker version 18.06.0-ce, build 0ffa825