* Two Pass Noise Reduction Question (newbie)

rtbond wrote on 6/7/2003, 11:58 AM
Hello,

I'm experimenting with the SoFo Noise Reduction 2 plugin. The documentation suggests trying a two pass noise reduction process for "stubborn" noise. That is, running the source material through the NR2 plugin, save the results, reload processed audio track, get a new noise print, and process a second time.

***What's the best audio format to be saving the file to after the first NR2 processing pass?***

Why the 2nd pass? Well after the first pass the noise is reduce, but still more noticeable tahn I'd like. The NR2 plugin is applying 12 dB of attenuation. I could bump up the attenuation from pass #1, but the manual seems to suggest doing a second pass is a better alternative.

Thanks!

--Rob

BTW: The noise is a buzz that I apparently picked up from routing my microphone cables too close to some stage lighting. The sound was clean during the sound check, but when the performance started (and this lighting fired up) I immediately noticed the buzzing in my headset. The buzzing is most noticeable in quiet passages, where I'm guessing the Digital8 camera's AGC is kicking in. I used a pair of Rode NT5 condenser mics, with phantom powering from a battery operated Rolls PB223, a Studio 1 XLR-BP Pro adapter (XLR to unbalanced mini-stereo).

Rob Bond

My System Info:

  • Vegas Pro 22 Build 194
  • OS: Windows 11.0 Home (64-bit), Version: 10.0.26100 Build 26100
  • Processor: i9-10940X CPU @ 3.30GHz (14 core)
  • Physical memory: 64GB (Corsair Vengeance LPX 64GB (2 x 32GB) DDR4 DRAM 3200MHz C16 memory kit)
  • Motherboard Model: MSI x299 Creator (MS-7B96)
  • GPU: EVGA GeForce RTX 2070 SUPER XC ULTRA (Studio Driver Version =  536.40)
  • Storage: Dual Samsung 970 EVO 1TB SSD (boot and Render); WDC WD4004FZWX, 7200 RPM (media)
  • Primary Display: Dell UltraSharp 27, U2723QE, 4K monitor with 98% DCI-P3 and DisplayHDR 400 with Dell Display Manager
  • Secondary Display: LG 32UK550-B, entry-level 4k/HDR-10 level monitor, @95% DCI-P3 coverage

Comments

Chienworks wrote on 6/7/2003, 12:11 PM
Save it as a .wav file with whatever format the sound came in from camera. Digital 8 is 16 bit stereo 48KHz, right? Every format change includes resampling which can degrade the audio, so keep the format consistant.
rtbond wrote on 6/7/2003, 7:26 PM
Thanks for your response regarding the file format.

Another related question is how long of a noiseprint (sample) should I look to use? Is a longer noise sample (e.g., 1 sec) better than a shorter one?

Thanks!

--Rob

Rob Bond

My System Info:

  • Vegas Pro 22 Build 194
  • OS: Windows 11.0 Home (64-bit), Version: 10.0.26100 Build 26100
  • Processor: i9-10940X CPU @ 3.30GHz (14 core)
  • Physical memory: 64GB (Corsair Vengeance LPX 64GB (2 x 32GB) DDR4 DRAM 3200MHz C16 memory kit)
  • Motherboard Model: MSI x299 Creator (MS-7B96)
  • GPU: EVGA GeForce RTX 2070 SUPER XC ULTRA (Studio Driver Version =  536.40)
  • Storage: Dual Samsung 970 EVO 1TB SSD (boot and Render); WDC WD4004FZWX, 7200 RPM (media)
  • Primary Display: Dell UltraSharp 27, U2723QE, 4K monitor with 98% DCI-P3 and DisplayHDR 400 with Dell Display Manager
  • Secondary Display: LG 32UK550-B, entry-level 4k/HDR-10 level monitor, @95% DCI-P3 coverage
Chienworks wrote on 6/7/2003, 9:57 PM
You need to grab a representative sample of the sound. Most noise is relatively random so even a short section of it will do. For 60Hz hum grabbing a a few 60th's of a second will do. If the sound has a slower cycle (for example, a noise that rises and falls every few seconds) then you'll have to grab enough to contain at least one full cycle to be effective.
LarryP wrote on 6/7/2003, 10:41 PM
Try taking a noise print and setting perhaps 20db of noise reduction. Save this as a preset (the little floppy disk). Insert a second NR2 filter in the audio chain and select the same preset.

I recently did this with great results to remove some low level hum (I hate unbalanced audio!). By doing this I was able to use mode 1 instead of mode 0 which resulted in fewer artifacts.
rtbond wrote on 6/8/2003, 2:06 PM
Larry,

Is there any benefit to providing a different noise print to the second NR2 in the noise reduction process (i.e., a noise print of what remains after the 1st NR2 pass?).

I assume that by using the same preset on the second pass you are effectively using the same noise print on the second NR2 in the chain as the 1st NR2.

Thanks!

--Rob

Rob Bond

My System Info:

  • Vegas Pro 22 Build 194
  • OS: Windows 11.0 Home (64-bit), Version: 10.0.26100 Build 26100
  • Processor: i9-10940X CPU @ 3.30GHz (14 core)
  • Physical memory: 64GB (Corsair Vengeance LPX 64GB (2 x 32GB) DDR4 DRAM 3200MHz C16 memory kit)
  • Motherboard Model: MSI x299 Creator (MS-7B96)
  • GPU: EVGA GeForce RTX 2070 SUPER XC ULTRA (Studio Driver Version =  536.40)
  • Storage: Dual Samsung 970 EVO 1TB SSD (boot and Render); WDC WD4004FZWX, 7200 RPM (media)
  • Primary Display: Dell UltraSharp 27, U2723QE, 4K monitor with 98% DCI-P3 and DisplayHDR 400 with Dell Display Manager
  • Secondary Display: LG 32UK550-B, entry-level 4k/HDR-10 level monitor, @95% DCI-P3 coverage
seeker wrote on 6/8/2003, 8:44 PM
Rob,

Pardon me for jumping in, but this thread is quite interesting to me because I am trying to learn the ins and outs of NR2 myself. The Noise Reduction 2 manual leaves a lot of questions unanswered.

"I assume that by using the same preset on the second pass you are effectively using the same noise print on the second NR2 in the chain as the 1st NR2."

I think Larry will say "Yes" to that. Actually, I think Larry has suggested a very good technique there. Doing two 20db noise reductions is not the same as doing one 40db or 30db noise reduction, but is potentially better. I plan to try Larry's suggestion myself.

"Is there any benefit to providing a different noise print to the second NR2 in the noise reduction process (i.e., a noise print of what remains after the 1st NR2 pass?)."

I am curious what Larry will say about that question. In Vegas I am thinking about doing event-level rather than track-level noise reduction. My captured sound usually has several "silent" spots suitable for taking a noise sample, and there is no reason to believe that video taken "in the field" has uniform noise from one clip to the next. And since I am using an on-camera shotgun mic, my noise could shift even during a pan. So I am faced with the conundrum of wanting to take multiple noise samples but being a bit uncertain of how best to use them. There needs to be a whole book about how to use Noise Reduction 2.

-- Seeker --
LarryP wrote on 6/8/2003, 9:43 PM
About using the same noise profile my thinking was that the noise doesn't change from NR to NR. If you take a noise profile from the output of the first filter you are profiling the noise plus any artifacts introduced by the first filter instead of just the noise. My little test is not very scientific but I am happy with the result.

Interesting about applying an FX to an event. I was looking yesterday for a way to do so. In the video tracks there is a little FX icon. The only way, maybe, is to use the non real time effects. Anybody have any ideas?

I just tried the 2 filters and preset to get rid of motor noise on my Canon ZR65 and the results were very good.

Larry