Add NR2 (emnr) NR algo from WDSP library

[grahamwhaley]

It looks like I 'missed' an NR algo in my original research. I added the xanr algo from WDSP, but completely missed that it also contains a spectral NR algo, generally called NR2 (as that is how it is labelled in the relevant GUIs), but can be found in the file emnr.c in the library itself.

It could be quite a challenge to integrate, as:

• it seems to have many variables and options, although there are recommended defaults
• it uses a 'calculus' datafile, which nicely in the g0orx wdsp port it has been converted to a C array datafile, but it is ~1Mb large, so we need to see/check if we can build that for the teensy (I expect so), and if it will fit in the flash! The Teensy 4.1 has 8Mb of flash, so I think we should be OK!

[Roturbo]

Hello @grahamwhaley.

I have upload 2 videos to YouTube as an example of difference from NR to NR2 on Thetis software.

https://youtu.be/nLl0pLms4EM

https://youtu.be/AWhHCL32htw

I was looking for the Voice Squelch, but now i think the NR2 must come first,
i hope you have some time one day to make upgrades on your DSPham.

Regards

[grahamwhaley]

Hi @Roturbo - thanks for the videos! NR2 is sounding pretty good in those. NR2 is closer to Spectral in DSPham than to NR. DSPham already has NR, but it is called LLMS in DSPham.

I also realised we are normally using Teensy 4.0 in DSPham, not Teensy 4.1 (although a 4.1 should work fine I think). But Teensy 4.0 'only' has 2Mb of Flash, so we would have to see if NR2 would fit!

Thetis/PowerSDR is looking very nice in those videos btw - but, I think I can't run it under Linux with any of my radios or SDR rx's :-)

[BitLoose]

Hi, sounds great on the video, wan't sure which was which though as both versions NR1 and 2 sound quite good, the first seemed clearer though. 73, Rob

…

On 08/06/2023 03:14, Roturbo wrote: Hello @grahamwhaley <https://github.com/grahamwhaley>. I have upload a video to youtube as an example of difference from NR to NR2. https://youtu.be/nLl0pLms4EM I was looking for the Voice Squelch, but now i think the NR2 must come first, i hope you have some time one day to make some upgrades on your DSPham. Regards — Reply to this email directly, view it on GitHub <#28 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADYZGDYWZMZ4WTAC2VKWVJDXKEYQZANCNFSM6AAAAAAY5URY6Y>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[Roturbo]

Hello @grahamwhaley

Yes the first time i switch from DSPham to this software radio it sounds almost the same
and there were moments the DSPham sounds better, but it is a question of get use to it.

About Thetis, it works only with the Apache Labs radios and Hemes boards,
there is a linux version at https://github.com/g0orx/pihpsdr again for the same radios.

I have test the PiHPSDR and is more silence without NR´s than the win version, still,
the linux version runs also on Raspberry and is not updated for long time.

Regards

[Roturbo]

Hello @BitLoose, on NR there is always some noise on the back, also the audio is not clear as
with NR2 https://youtu.be/nLl0pLms4EM on this video watch on minute 2:13.

You will notice the the transition from NR2 to normal,, than i switch to NR a few seconds
and there is the background noise,, after on NR2 again there is no noise and the voice
is more clear and loud.

Regards