Saturday, November 30, 2024

AI noise removal from off air single sideband

The applications of AI (machine learning) to radio are fascinating me at the moment. This is a simple experiment where I recorded some SSB and fed it in to an AI noise remover at https://audo.ai/noise-removal This video switches between the input and the resulting output.


This was done just with the default settings. No training on just the noise was done. I think it does a very good job. My guess is that with modern CPUs and ML chips it should be possible to do this sort of processing in near to real time.

 

2 comments:

Bill Meara said...

It is impressive Peter. This may end up being a generational thing: I kind of like the noise (within limits, of course) because it lets us know if our receivers are working or not. There is the old test of receiver sensitivity: connect the antenna -- if the noise level increases the receiver is sensitive enough. But it seems that younger hams see the noise as a bug, not a feature. They see it as a defect, something to be eliminated. I even heard one guy on the air talking about a scheme in which received signals would be shipped off via the internet for de-noise-ification, and then shipped back in time for a (noiseless) listen and response. To me, that seems like a lot of trouble to go through. 73 Bill N2CQR

Peter Marks said...

Thanks Bill, I'm looking to the future where this could be done in real time like Noise Reduction NR is done now. I would listen to the unprocessed signal first for sure like you. 73 Peter VK3TPM