RF signals and AI studies applied to Signal Processing for Amatuer Radio Applications Including Morse Code
Windows and Raspberry Pi are being used to develop python code and test sound/audio capture from SDR and analog HAM radios using inexpensive USB sound dongles. The code is being written and developed to learn about AI applications for signal processing. A textbook Deep Learning with PyTorch, Manning Press, is being studied with this effort.
Initial examples were to generate a visual pattern for morse code signals based on indivisual letters, numbers, and symbols as well as small common communication strings. Several python modules for sound have been tried including pygame, sounddevice, and pyaudio. The pyqtgraph and matplotlib graphing modules, and the scipy and numpy math array packages are heavily used. Some of the code has been compared to real CW signals captured using the Audacity program to record audio and convert to a spectrogram. CW_spectrogramv1 displays a CW audio signal in near realtime using a python script. Hopefully the images can be used to feed the Deep Learning algorithms. The initial assumption was to work directly with raw audio. Further study seems to imply that conversion to a frequency domain and allowing multiple line output of communicaiton might be a better approach.
This author wants to acknowledge and thank the many contributors to github and python articles. The code written so far heavily borrows from previous work but hopefully the focus will lead to something new.