[updated 18 August 2024]

KiwiClient

This is a Python client for KiwiSDR. It allows you to:

• Receive data streams with audio samples, IQ samples, and waterfall data
• Issue commands to the KiwiSDR

Install

Visit github.com/jks-prv/kiwiclient
Click the green 'Code' button and select 'Download ZIP'
You should find the file 'kiwiclient-master.zip' in your download directory.
Copy it to the appropriate destination and unzip it if necessary.
Perhaps rename the resulting directory 'kiwiclient-master' to just 'kiwiclient'.
Change to that directory.
Here you will find a 'Makefile' containing various usage examples.
Assuming your system has 'Make' and 'Python3' installed type 'make help' to get started.
Or without 'Make' type 'python3 kiwirecorder.py --help'
It is also possible to use the 'git' tools to checkout a kiwiclient clone that is easier to keep updated.

Dependencies

It is strongly recommended you use Python3.

Make sure the Python package 'numpy' is installed.
On many Linux distributions the command would be similar to 'apt install python3-numpy'
On macOS try 'pip3 install numpy' or perhaps 'python3 -m pip install numpy'

Resampling

If you want high-quality resampling based on libsamplerate (SRC) you should build the version included with KiwiClient that has fixes rather than using the standard python-samplerate package.
Follow these steps. Ask on the Kiwi forum if you have problems: 'forum.kiwisdr.com'

• Install the Python package 'cffi'
• Install the 'libsamplerate' library using your system's package manager. Note: this is not the Python package 'samplerate' but the native code library 'libsamplerate' (e.g. x86_64 or arm64).
  • Windows: download from 'github.com/libsndfile/libsamplerate/releases'
  • Linux: use a package manager, e.g. 'apt install libsamplerate' or 'libsamplerate0'
  • macOS: use a package manager like brew: 'brew install libsamplerate'
• Run the samplerate module builder 'make samplerate_build'. This generates a Python wrapper around 'libsamplerate' in the file 'samplerate/_src.py'
• Install 'pytest' using the Python package manager or perhaps 'pip3 install pytest'
• Test by running 'make samplerate_test'
• If your system says 'pytest' isn't found try 'make samplerate_test2'

If you can't build the Kiwi version then install the regular Python package: 'pip3 install samplerate'
If either samplerate module is not found then low-quality resampling based on linear interpolation is used.
A message indicating which resampler is being used will be shown.
Set the environment variable 'USE_LIBSAMPLERATE' to 'False' to force the linear interpolator to be used.

Demo code

The following demo programs are provided. Use the --help argument to see all program options.

• kiwirecorder: Record audio to WAV files, with squelch. Option --wf prints various waterfall statistics.
  Adding option --wf-png records the waterfall as a PNG file. --help for more info.
• kiwiwfrecorder: Specialty program. Saves waterfall data and GPS timestamps to .npy format file.
• kiwifax: Decode radiofax and save as PNGs, with auto start, stop, and phasing.
• kiwiclientd: Plays Kiwi audio on sound cards (real & virtual) for use by programs like fldigi and wsjtx. Implements hamlib rigctl network interface so the Kiwi freq & mode can be controlled by these programs.
• kiwi_nc: Deprecated. Use the --nc option with kiwirecorder. Command line pipeline tool in the style of netcat. Example: streaming IQ samples to dumphfdl (see the Makefile target dumphfdl).

The Makefile contains numerous examples of how to use these programs.

IS0KYB micro tools

Two utilities have been added to simplify the waterfall data acquisition/storage and data analysis. The SNR ratio (a la Pierre Ynard) is computed each time. There is now the possibility to change zoom level and offset frequency.

• microkiwi_waterfall.py: launch this program with no filename and just the SNR will be computed, with a filename, the raw waterfall data is saved. Launch with --help to list all options.
• waterfall_data_analysis.ipynb: this is a demo jupyther notebook to interactively analyze waterfall data. Easily transformable into a standalone python program.

The data is, at the moment, transferred in uncompressed format.

Guide to the code

kiwiclient.py

Base class for receiving websocket data from a KiwiSDR. It provides the following methods which can be used in derived classes:

• _process_audio_samples(self, seq, samples, rssi): audio samples
• _process_iq_samples(self, seq, samples, rssi, gps): IQ samples
• _process_waterfall_samples(self, seq, samples): waterfall data

kiwirecorder.py

• Can record audio data, IQ samples, and waterfall data.
• The complete list of options can be obtained by python3 kiwirecorder.py --help.
• It is possible to record from more than one KiwiSDR simultaneously, see again --help.
• For recording IQ samples there is the -w or --kiwi-wav option: this writes a .wav file which includes GNSS timestamps (see below).
• AGC options can be specified in a YAML-formatted file, --agc-yaml option, see default_agc.yaml. Note that this option needs PyYAML to be installed

IQ .wav files with GNSS timestamps

kiwirecorder.py configuration

• Use the option -m iq --kiwi-wav --station=[name] for recording IQ samples with GNSS time stamps.
• The resulting .wav files contains non-standard WAV chunks with GNSS timestamps.
• If a directory with name gnss_pos/ exists, a text file gnss_pos/[name].txt will be created which contains latitude and longitude as provided by the KiwiSDR; existing files are overwritten.

Working with the recorded .wav files

• There is an octave extension for reading such WAV files, see read_kiwi_wav.cc where the details of the non-standard WAV chunk can be found; it needs to be compiled in this way: mkoctfile read_kiwi_wav.cc.
• For using read_kiwi_wav an octave function proc_kiwi_iq_wav.m is provided; type help proc_kiwi_iq_wav in octave for documentation.

[end-of-document]