Great... Working as designed.
AL, K0VM
On 5/14/2020 10:18 AM, Bill Somerville wrote:
On 14/05/2020 15:55, Al wrote:
Just FWIW...
I have set up a test where I have 2.2.0-RC1 and 2.1.2 running in
parallel on the same PC from the same audio source. I notice that the
SNR reported by 2.2.0-RC1 is consistently more optimistic vs.
2.1.2. In some cases up to 13db greater on strong signals. Most
signals report at least a few db greater.
Just an observation..
AL, K0VM
Hi Al,
it's no so much optimistic as accurate. v2.2.0 introduces a window
filter on the FFT that is used for noise estimation, it greatly
reduces bias from strong signals in the pass band. You should notice
that the SNR of strong signals is not compressed anywhere near as much
as previous versions, e..g. you may see SNR values above +20 being far
more accurate and they may extend up to +50 or more.
73
Bill
G4WJS.
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
