Added another readme

README_our.md

@@ -0,0 +1,41 @@
+== Installation ==
+
+Installation is simple - just install tensorflow 1.4.1:
+
+python setup.py -r requirements.txt
+
+for training you need a GPU version, decoding can be done on CPU
+
+== Data preparation ==
+
+To prepare the data put wav files into data folder.
+
+The files must be WAV PCM data 44100 Hz mono
+
+The original and processed files should have the same filename, just differ by the path:
+
+data/ENG/processed/1679/UK_06.PR 001_01INTRO_001.wav
+data/ENG/original/1679/UK_06.PR 001_01INTRO_001.wav
+
+The system uses the filename to identify which files converted to which, so filename is important
+
+Put the list of original files into the file data/wav.list, the system will take data from this file, one file per line.
+
+== Training ==
+
+Training is two steps, first prepare the data and convert it to TF format:
+
+./prepare_data.sh
+
+Then train the model
+
+./train_segan.sh
+
+Training will take some time
+
+== Cleaning the files ==
+
+To clean the data use simple clean_wav.sh script:
+
+./clean_wav.sh noisy.wav clean.wav
+