Re: [Elecraft] K3 noise reduction revisited
Great news Lyle, hope to see some definite improvements in the next few firmware releases. 73 de James K2QI On Mon, Jul 13, 2009 at 3:22 PM, Lyle Johnson wrote: > > Has there been any progress on improving the NR algorithm used in the K3? > > A lot of effort is going into this at the moment and we hope to > something ready soon. > > 73, > > Lyle kK7P > > __ > Elecraft mailing list > Home: http://mailman.qth.net/mailman/listinfo/elecraft > Help: http://mailman.qth.net/mmfaq.htm > Post: mailto:Elecraft@mailman.qth.net > > This list hosted by: http://www.qsl.net > Please help support this email list: http://www.qsl.net/donate.html > __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 noise reduction revisited
> Has there been any progress on improving the NR algorithm used in the K3? A lot of effort is going into this at the moment and we hope to something ready soon. 73, Lyle kK7P __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 noise reduction revisited
Nevermind; found the AFX button and was able to turn it off completely. Unfortunately, still no major improvement in the NR performance. James K2QI On Sun, Jul 12, 2009 at 4:53 PM, K2QI wrote: > Rick, > > I'm assuming that the 'BIN' setting is what turns the AFX off? I've tried > the AFX settings with delay going from 1-5 plus BIN, and there isn't too > much of a noticeable difference. The DSP still works poorly. It probaly > wouldn't be so bad if the DSP didn't attenuate the signal as much as it does > also. > > I regret selling my BHI ANEM module now; I could have used it line with my > headphones. > > James K2QI > > On Sun, Jul 12, 2009 at 3:40 PM, Rick Braun wrote: > >> I posted this once before, but not a single response, so here goes >> again. I found that the F1-1 NR works OK if the AFX is turned off. My >> sense is that AFX plus NR is too much processing. Try the NR without the >> fake stereo, it seems more effective to me. >> >> Rick Braun >> KE7VXP >> K3 # 2869 >> >> - Original Message ----- >> *From:* James Sarte >> *To:* elecraft@mailman.qth.net >> *Sent:* Saturday, July 11, 2009 9:06 PM >> *Subject:* [Elecraft] K3 noise reduction revisited >> >> Has there been any progress on improving the NR algorithm used in the K3? >> As it is, it's only helpful perhaps 2% of the time on SSB with the >> remaining >> 98% of it remaining off. I've tried adjusting AGC settings, plus fiddling >> with the DSP settings and can never seem to find a happy medium. >> >> >> >> Adjusting the receive EQ also has a major impact on the NR's audio >> quality. >> Sometimes I like to boost the midrange and treble areas on receive EQ as >> it >> helps with speech, but with any of the parameters set to a +1 or higher, >> the >> resulting audio out with NR on is terrible. It sounds almost as if I'm >> listening to the audio through a long drain pipe. >> >> >> >> Finally, turning on the NR generally tends to attenuate the signal rather >> than filter it; a very annoying behavior if you ask me. This has become >> painfully obvious during this weekend's IARU contest. >> >> >> >> Truth be told, the best DSP application I've used and heard so far has >> come >> from BHI with their ANEM modules. There are only a few settings to >> adjust, >> but it's very effective in isolating human speech patterns from background >> noise. It runs rings around the K3 when it comes to SSB noise reduction. >> I'm hoping that sooner (rather than later), Elecraft can come up with a >> firmware fix for the NR. >> >> >> >> 73 de James K2QI >> >> __ >> Elecraft mailing list >> Home: http://mailman.qth.net/mailman/listinfo/elecraft >> Help: http://mailman.qth.net/mmfaq.htm >> Post: mailto:Elecraft@mailman.qth.net >> >> This list hosted by: http://www.qsl.net >> Please help support this email list: http://www.qsl.net/donate.html >> >> > __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
[Elecraft] K3 noise reduction revisited
Has there been any progress on improving the NR algorithm used in the K3? As it is, it's only helpful perhaps 2% of the time on SSB with the remaining 98% of it remaining off. I've tried adjusting AGC settings, plus fiddling with the DSP settings and can never seem to find a happy medium. Adjusting the receive EQ also has a major impact on the NR's audio quality. Sometimes I like to boost the midrange and treble areas on receive EQ as it helps with speech, but with any of the parameters set to a +1 or higher, the resulting audio out with NR on is terrible. It sounds almost as if I'm listening to the audio through a long drain pipe. Finally, turning on the NR generally tends to attenuate the signal rather than filter it; a very annoying behavior if you ask me. This has become painfully obvious during this weekend's IARU contest. Truth be told, the best DSP application I've used and heard so far has come from BHI with their ANEM modules. There are only a few settings to adjust, but it's very effective in isolating human speech patterns from background noise. It runs rings around the K3 when it comes to SSB noise reduction. I'm hoping that sooner (rather than later), Elecraft can come up with a firmware fix for the NR. 73 de James K2QI __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 Noise Reduction..revisited
k6...@foothill.net wrote: > They will notch a fire truck siren down to a tolerable > level I don't know if I like that or not ;-) > However, they apparently think CW is a fire truck. > I can safely say that for the many years I have on the job, CW is definitely not a fire truck :-) Drivers would probably get out of the way for CW ;-) 73, Dave W8FGU __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 Noise Reduction..revisited
Don Wilhelm wrote: > > Those Noise Reduction algorithms you refer to seem to be quite clever > and effective. They appear to be close to the NR that I have in my > hearing aids which does a great job of extracting speech from noise. > Unfortunately, the NR in my hearing aids also thinks sustained musical > notes are noise, and by the same token, CW can also be considered noise. My experience also. The VA gave me some new digital [DSP] hearing aids a few years ago. They will notch a fire truck siren down to a tolerable level, and do a fine job of extracting voice and music from noise [except for sustained notes]. I hadn't heard birds in nearly 40 years until I got these. However, they apparently think CW is a fire truck. I am finding that my K3 NR performance is very sensitive to other RX settings as well as the DSP BW. The recently posted suggestions for AGC settings have helped a lot. I'm pretty much CW-only because of my hearing, I have the stock roofing filters, and I usually run the DSP around 250 Hz. Engaging NR at one of the lesser aggressive settings raises the volume of the signal, I turn down the AF Gain, and the noise subsides nicely with the signal now standing out in my headphones better. There seems to be a threshold level below which a very weak signal can't be found by the NR algorithm but I haven't had time to experiment with the NR settings yet. Possibly a more agressive setting will lower that threshold. Really agressive settings seem to create more garbage than they reduce noise. Haven't gotten to SSB yet. My K3 is definitely NOT your father's radio :-) 73, Fred K6DGW - Northern California Contest Club - Auburn CA [Placer County, CM98LW] __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 Noise Reduction..revisited
Lyle is looking at the NR code. Let's rest this thread for now. We'll pop up here when we have something new to try. :-) 73, Eric WA6HHQ __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 Noise Reduction..revisited
Regarding noise reduction and the BHI products; I've noticed that the BHI ANEM works substantially better than the K3's NR. For whatever reason, whenever I switch NR on, the band noise seems to get attenuated slightly, but then reshaped to resemble noise in a tunnel. It's very tiring to listen too. Any suggestions? 73 de James K3JPS Trevor Smithers wrote: > As far as I'm aware there are two companies that produce this technology for the amateur > market, BHI in the UK and Michels-Engineering in Germany. The german unit is designed > specifically for the extraction of the speech waveform when surrounded by noise. BHI is similar > but works with cw as well. > > Here are a couple of links to audio samples of both units in operation > http://www.ing-michels.de/audio_demonstrations.html > http://www.radio.bhinstrumentation.co.uk/html/demonstration.html > > There is a downside to using these units in that they only work with a mono input so you loose the > binaural function. To get around this I purchased two of the internal modules (DX21) from > Michels-Engineering and wired them to each stereo channel. This allows independent control of > the noise reduction in each ear and retains the AFX/binaural function. > > 73 > Trevor G0KTN __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 Noise Reduction..revisited
Hmm, are Lyle's "other irons in the fire" being heated for Eric's "other fish to fry"? :-) More seriously, I also would love to see changes to the noise reduction, particularly to help with weak signal work. I'm not sure why anyone cares about noise reduction for strong signals -- I just turn down the AF gain and the noise goes away :-) Steve N9SZ Don W3FPR wrote: Perhaps Lyle is looking at other alternatives for NR in the K3, but I do know that he is quite busy with other 'irons in the fire' right now, so patience is a virtue for now and we could possibly discover improvements down the road. __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 Noise Reduction..revisited
Trevor, Those Noise Reduction algorithms you refer to seem to be quite clever and effective. They appear to be close to the NR that I have in my hearing aids which does a great job of extracting speech from noise. Unfortunately, the NR in my hearing aids also thinks sustained musical notes are noise, and by the same token, CW can also be considered noise. Nothing is perfect, and compromises must be made, but the current NR algorithm used in the K3 does not work for me either - it causes severe distortion of SSB signals (speech) and while my AGC settings do not create the wild signal increases on CW that others have reported, I usually find that narrow filtering is more pleasant to use than the NR function. Perhaps Lyle is looking at other alternatives for NR in the K3, but I do know that he is quite busy with other 'irons in the fire' right now, so patience is a virtue for now and we could possibly discover improvements down the road. All these thoughts are entirely my own and are based on pure speculation - I have no further information than what has been posted to the Elecraft reflector. 73, Don W3FPR Trevor Smithers wrote: > As far as I'm aware there are two companies that produce this technology for > the amateur > market, BHI in the UK and Michels-Engineering in Germany. The german unit is > designed > specifically for the extraction of the speech waveform when surrounded by > noise. BHI is similar > but works with cw as well. > > Here are a couple of links to audio samples of both units in operation > http://www.ing-michels.de/audio_demonstrations.html > http://www.radio.bhinstrumentation.co.uk/html/demonstration.html > > There is a downside to using these units in that they only work with a mono > input so you loose the > binaural function. To get around this I purchased two of the internal modules > (DX21) from > Michels-Engineering and wired them to each stereo channel. This allows > independent control of > the noise reduction in each ear and retains the AFX/binaural function. > > 73 > Trevor G0KTN > > > __ > Elecraft mailing list > Home: http://mailman.qth.net/mailman/listinfo/elecraft > Help: http://mailman.qth.net/mmfaq.htm > Post: mailto:Elecraft@mailman.qth.net > > This list hosted by: http://www.qsl.net > Please help support this email list: http://www.qsl.net/donate.html > > > > No virus found in this incoming message. > Checked by AVG - www.avg.com > Version: 8.5.287 / Virus Database: 270.12.8/2086 - Release Date: 04/29/09 > 06:37:00 > > __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 Noise Reduction..revisited
As far as I'm aware there are two companies that produce this technology for the amateur market, BHI in the UK and Michels-Engineering in Germany. The german unit is designed specifically for the extraction of the speech waveform when surrounded by noise. BHI is similar but works with cw as well. Here are a couple of links to audio samples of both units in operation http://www.ing-michels.de/audio_demonstrations.html http://www.radio.bhinstrumentation.co.uk/html/demonstration.html There is a downside to using these units in that they only work with a mono input so you loose the binaural function. To get around this I purchased two of the internal modules (DX21) from Michels-Engineering and wired them to each stereo channel. This allows independent control of the noise reduction in each ear and retains the AFX/binaural function. 73 Trevor G0KTN __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 Noise Reduction..revisited
I agree, whatever algorithm BHI uses, it really works. 73 Stewart G3RXQ On Wed, 29 Apr 2009 16:05:07 -0500, Bill NY9H wrote: > I have two "Hear-it" speakers from England's BHI company hooked thru > my microham controller's headphone outputwhich is connected to > the 7800 & the K3. > > With a low- mid aggressive setting they don't sound anything but > quieter. (what a concept)... leave them in line most all the time. > > No gurgling, whooshing... or D-Star audio emulations. > (Much better than the NCT licensed stuff Clearspeech/Heil now West > Mountain Radio). > > Granted if I was using a GrandioseSweetness18" 3 way, I might have a > different outcome. Then again I could pipe it thru my UREI > "time-aligned" monitors. > > 40 years in the audio business.tells me the BHI algorithm is a > winner whatever they are doing ...I wish Lyle would do ( if the > DSP can handle it ) > And I do know the existing design does it the other way . > > Seems we revisit this one about every 118 days.. > > bill > > __ > Elecraft mailing list > Home: http://mailman.qth.net/mailman/listinfo/elecraft > Help: http://mailman.qth.net/mmfaq.htm > Post: mailto:Elecraft@mailman.qth.net > > This list hosted by: http://www.qsl.net > Please help support this email list: http://www.qsl.net/donate.html __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
Re: [Elecraft] K3 Noise Reduction..revisited
I have two "Hear-it" speakers from England's BHI company hooked thru my microham controller's headphone outputwhich is connected to the 7800 & the K3. With a low- mid aggressive setting they don't sound anything but quieter. (what a concept)... leave them in line most all the time. No gurgling, whooshing... or D-Star audio emulations. (Much better than the NCT licensed stuff Clearspeech/Heil now West Mountain Radio). Granted if I was using a GrandioseSweetness18" 3 way, I might have a different outcome. Then again I could pipe it thru my UREI "time-aligned" monitors. 40 years in the audio business.tells me the BHI algorithm is a winner whatever they are doing ...I wish Lyle would do ( if the DSP can handle it ) And I do know the existing design does it the other way . Seems we revisit this one about every 118 days.. bill __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
