I talked about the FreeDV project and demonstrated one of the best performing modes, 700D with its hand crafted (by David, VK5DGR) codec and modulation and then compared it with the new Radio Autoencoder (RADE) which uses machine learning to transmit amazing sounding voice over an RF bandwidth of about 1.5kHz.
There are many applications of machine learning in the amateur radio world that come to mind including:
- Morse decoding
- Noise removal
- Speech transcription
- More digital modes
We live at an amazing time in technology evolution.
2 comments:
The emergence of AI LLMs has changed my life... as well as using LLMs for finding and explaining all sorts of information, I am relying on it in my very part time hobby of designing PCBs for radio projects. At the moment I am usig it to design a small hand-held CQ transceiver, and Gemini is incredibly helpful in choosing components, interpreting data sheets, advising on design options such as biasing, designing filters, configuring signal levels around mixers, validating substitutes, the list goes on. Questions that previously would have held me up for days are now answered in seconds. I know that LLMs are not always correct. So I try to validate everything independently. The proof of my AI RF design engineer assistant will be in whether my desgn works! Stay tuned.
Yes, amazing. I'm using Claude and it did most of the work of writing a Google takeout to Wordpress migrator - not perfect but a fantastic start.
Post a Comment