Just a quick side project

---

There are 2 scripts, one is a re-written one in Python 3 morse-to-text-python3.py and the other one is written by me from scratch.

morse-to-text-python3.py

This script has been based on some Python 2 code I found searching for this decoder, it was found on stackoverflow post mentioning this link.

I haven't fully investigated the workings, but here is the principals.

1. Takes in a .wav file
2. Runs a Fourier Transform on it numpy.fft.rfft numpy functions

• I think this is for cleaning up the audio

3. Looks for pulses and generates a plot
4. Translates the pulses to Morse code
5. Then using the codes.csv file, translates the Morse code

It generates a plot after every step of the process and you can see all of them in the report.html.

I couldn't understand that much of it, but I at least remade it in Python 3 (although I could have started at first from scratch), either it was too complicated for me or poorly written. (Not that I think I did better.)

The original file is morse-to-text.py, it includes usage with command line arguments (mine remake doesn't).

morse-audio-decoder.py

This script has 2 feature, one is a decoder, the other one is an unnecessary one, noise reduction feature, that uses sox *nix utility.

Usage

usage: morse-audio-decoder.py [-h] [-l LEVEL] {clean,decode} ... file

positional arguments:
  {clean,decode}        Commands for the decoder
    clean               Clean up the audio
    decode              Decode morse code
  file                  Input file name

optional arguments:
  -h, --help            show this help message and exit
  -l LEVEL, --level LEVEL
                        Verbosity level

Morse Audio Decoder

This feature takes in a .wav file and does the following.

1. With pydub transforms it into an audio segment
2. Coverts all the values that are above the max - (max * max_threshhold) to 1 and others to 0
3. Iterated through all the binary values, when the value switches to a different one:

• Length is calculated in milliseconds and then based on the length, the morse code symbol is determined.

DIH = 100  # On default and the same length for symbol space
DAH = DIH * 3  # The same length is assigned for letter spacing space
SPACE = DIH * 7  # The space between words

4. Reads the codes.csv file and assigns it to a dictionary.
5. Finally the characters separated by the Character space are compared against the dictionary and translated

• Unknown characters are assigned ?
• Word spaces are translated to a space and letter space is assigned to an empty string

Options

-d | --dih - The length of DIH symbol in ms (will implement a detection for this) \

• 100ms is good enough for 15 WPM (tested for PARIS)
• 50-60ms - 20 WPM
• 25ms can do 50 WPM

-l | --level - The verbosity of the output, haven't done much thinking with this, but 0 is for all output.

Audio De-noiser

I thought I needed this feature, but will maybe improve it later, haven't tested it much, but all it does is

1. Takes a .wav file
2. Finds all 'silence' segments what are passed a given threshold ('-33 dBFS' on default)
3. Runs sox on every segment to create a noise profile
4. Passes the input file through all of the generated profiles
5. Exports a cleaned version of the audio file in export directory

To Do

• Improve the code for better detection
• Add an encoder from text to sound
• Add comparison option for testing
• Add Dynamic WPM detection (so you don't have to change -d manually)

Resources

• Good online decoder. also explains how it works.
• Morse code generator that I used for generating some samples.
• In depth about Morse code