The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Is white noise supposed to have a perfectly flat frequency response? Virtual Instrument Plugins
Old 2nd September 2011
  #1
Here for the gear
 

Is white noise supposed to have a perfectly flat frequency response?

it’s just that i have got the demo of Pinguin audio meter today and i had to send my stereo output from my soundcard back into the line in to get it to read anything seems as it won't pick a signal up from the output alone. The cables I’m using are pretty high quality, but I’m a bit concerned that pinguin might be seeing a coloured frequency range or seeing interference.

i tried testing it by loading up Omnisphere and generating some white noise, i expected it to be a perfectly flat response but instead it’s a bit uneven.

Does this look normal to you? Maybe if anyone else had pinguin have you tested or know of a good way to test that it's working correctly?

it's showing up with a small increase in the low and high end, and with a dip around the 200hz mark, with some spikes here and there (450hz)

what do you think?


Old 2nd September 2011
  #2
Lives for gear
 
Alexey Lukin's Avatar
 

Since noise is a random signal, its spectrum on a certain time segment is random too. To get more reliable estimates, you need to average spectra on multiple segments. Spectrum analyzers often have settings for averaging mode. Here's an example of white noise spectrum in RX:



Here the blue curve corresponds to an "instantaneous" spectrum (inside a single analysis window), while yellow curves correspond to different amounts of averaging (when you make a longer selection in RX, it automatically averages spectrum in it).
Old 2nd September 2011
  #3
Here for the gear
 

Thanks for the reply


i guess using white noise is not the right way to go.
I just tried a different test, i created a string of small sine wave blips, all exactly the same length and same velocity in Cubase, so they should all output the same volume.


for every one of the yellow peak dashes shown on the Pinguin meter to the right, that’s pretty much where one sine wave note was, for this analyser to be reading correctly i would have expected this one to be dead straight unless I’m not taking something into consideration. Big pic incomming!

Using sine waves am i right in thinking this should be picking up each note at the same level and so appearing as a straight line rather than having the up and down peaks?


Old 2nd September 2011
  #4
Here for the gear
 

I think i answered my own question i tried this test again in Voxengo SPAN, this is exactly as i had imagined it would show up.


But now I’m stuck because i don't know how I’m supposed to use pinguin, it just doesn't work unless i feed the output back in to my soundcard,


Does anyone else use pinguin audio meters and have this problem? how do you get around it? there doesn't seem to be any instructions or support on their site or anywhere on the net.



Old 2nd September 2011
  #5
Lives for gear
 
Alexey Lukin's Avatar
 

If you want to test your sound card, use RMAA. It's free and much more accurate than conventional spectrum analyzers:

Old 2nd September 2011
  #6
Registered User
 
Rick Sutton's Avatar
 

In reference to post #1.

Are you sure that you were using WHITE noise? IIRC WHITE noise is heavily weighted to the top with very little bottom. The graph looks more like a poorly generated pink noise.
Old 2nd September 2011
  #7
Here for the gear
 

Quote:
Originally Posted by Alexey Lukin View Post
If you want to test your sound card, use RMAA. It's free and much more accurate than conventional spectrum analyzers:
Thanks for the link! That is a cool little program to have! I tested it again with this and got the same results :( not sure why it’s so far off.



Quote:
Originally Posted by Rick Sutton View Post
In reference to post #1.

Are you sure that you were using WHITE noise? IIRC WHITE noise is heavily weighted to the top with very little bottom. The graph looks more like a poorly generated pink noise.
You’re correct! well spotted it was pink noise after all…doh! i just got a sample of white noise to test it again and it is more like how you described although you can still see the odd spikes where the signal is being coloured,


might need to make a new thread for some instructions on how to use this software correctly, cause it does really appeal to me, its convenient to have it running on another screen to the side of me so I can see it all the time when I need it, and its much clearer to see what is happening than most analysers I’ve used.

Old 2nd September 2011
  #8
restpause
Guest
Yeah I think you need well-generated pink noise, not white.
Old 2nd September 2011
  #9
Lives for gear
 
Alexey Lukin's Avatar
 

It depends on a spectrum analyzer type.
For linear-band analyzers based on FFT (such as Span and RX), horizontal line corresponds to white noise.
For log-band analyzers (such as Pinguin), horizontal line corresponds to pink noise.
Old 2nd September 2011
  #10
Lives for gear
 
valis's Avatar
As you've noticed, pinguin is using relatively fast responding fft bins in general, since you can't adjust the FFT size as you can in SPAN. Between that and the logarithmi response, you're just going to have to grow accustomed to monitoring pinguin in real world usages. I use it on an old P4 era laptop that is connected to 'studio' outs on my analog mixer that I don't use for anything else since I don't record bands in a separate 'studio' area, and find it very handy to get a quick handle on my spectral balance for whatever is audible on the board.

Also you should be aware that there's a "Pro" version of PG-AM which has linear as well as log responses (actually 5 responses) for the spectrum meter, as well as spectral content views and a few other added features. If memory serves it's also considerably more costly...
Old 2nd September 2011
  #11
Lives for gear
 
evosilica's Avatar
 

Quote:
Originally Posted by Rick Sutton View Post
In reference to post #1.

Are you sure that you were using WHITE noise? IIRC WHITE noise is heavily weighted to the top with very little bottom. The graph looks more like a poorly generated pink noise.
No! White noise has a flat frequency response. Pink has -20dB/Dec or -6dB/Oct and brown noise hase -40dB/Dec or -12dB/Oct.

If white noise looks weightet to the top end, then your spectrum analyzer is weighted (SPAN is by default weighted for example)

And this has nothing to do with log or lin band!
Old 2nd September 2011
  #12
Lives for gear
 
Alexey Lukin's Avatar
 

Quote:
Originally Posted by evosilica View Post
SPAN is by default weighted
Where did you get this information?


Quote:
Originally Posted by evosilica View Post
No! White noise has a flat frequency response. Pink has -20dB/Dec or -6dB/Oct and brown noise hase -40dB/Dec or -12dB/Oct. And this has nothing to do with log or lin band!
Sorry, you are wrong, just test a few analyzers to see that. Your quote is only true for a lin-band analyzer.
The definition of pink noise is that it has equal equal power in every octave, i.e. it will be a horizontal line on a log-band analyzer.
The definition of white noise is that it has equal power in every linear frequency interval, i.e. it will be a horizontal line on a lin-band analyzer (like FFT).

P.S. Noises do not have "frequency responses", you should operate in terms of power spectral density.
Old 5th September 2011
  #13
Lives for gear
 
evosilica's Avatar
 

Quote:
Originally Posted by Alexey Lukin View Post
Where did you get this information?
Check the edit menu. It is by default set to +6dB/oct and therefore makes pink noise more look like white noise.
EDIT: The parameter is called "slope" (probably in dB/oct) and it is set to +4.5 by default. It let's you tilt the whole spectrum. Set it to 0 to make it display accurately

Quote:
Originally Posted by Alexey Lukin View Post
Sorry, you are wrong
No I am not! heh

White noise = horizontal line on lin AND log analyzer
Pink noise = -6dB/oct Voltage or -3dB/Oct power on log analyzer and exponential shaped curve on a lin analyzer

White noise - Wikipedia, the free encyclopedia
Pink noise - Wikipedia, the free encyclopedia

LIN BAND
pink line: pink noise
blue line: white noise



LOG BAND
pink line: pink noise
blue line: white noise



Quote:
P.S. Noises do not have "frequency responses", you should operate in terms of power spectral density.
You mean signals in general. That's right.
Old 5th September 2011
  #14
Lives for gear
 
Alexey Lukin's Avatar
 

Quote:
Originally Posted by evosilica View Post
Check the edit menu. It is by default set to +6dB/oct and therefore makes pink noise more look like white noise.
EDIT: The parameter is called "slope" (probably in dB/oct) and it is set to +4.5 by default. It let's you tilt the whole spectrum. Set it to 0 to make it display accurately
Never used this analyzer, but thanks.


Quote:
Originally Posted by evosilica View Post
White noise = horizontal line on lin AND log analyzer
Sorry, you apparently don't understand what a log analyzer is, because both your graphs have actually been produced by the same lin-band analyzer. Although frequency scales are different, the power is still integrated by FFT in linear bands.
The example of a log-band analyzer is pretty much any hardware (like 1/3-octave) analyzer, e.g. this Klark Teknik. It integrates energy in log bands and will display pink noise as a horizontal line.
Old 5th September 2011
  #15
Lives for gear
 
evosilica's Avatar
 

I've just realized what i was confusing here.
You are refering to an analyzer that integrates the power within every band and since on a log scale the bands get larger and larger, the power also increases. So yeah, then pink noise becomes a horizontal line.

Sorry for the confusion, i was thinking about a linband analyzer with logarithmic visualisation .... but that's not a log band analyzer- Sorry
Old 5th September 2011
  #16
Registered User
 
Rick Sutton's Avatar
 

Just listen to white noise and pink noise and you'll see which one sounds like it represents all the frequencies from low to high equally to the ears and which one sounds like it is HEAVILY weighted to the top end (with so little low frequency information as to be difficult to find by the human ear). Then imagine how they would graph and remain correlated to how we hear them. To me pink noise sounds and graphs (on the analyzers I've used) like a full spectrum sound. White noise sounds and graphs like it is coming out of an old TV set.....hiss.
Of course they are related and are both useful, but to think of white noise as a flat spectrum only works in a mathematical scenario and bears very little relation to how we hear. At least how I hear!heh
Old 5th September 2011
  #17
Lives for gear
 
Alexey Lukin's Avatar
 

Critical bands of our hearing (which define ear's frequency integration properties) are approximately equally wide below 0.5 kHz and become proportionally wider above that. So, no surprise that pink noise saturates them equally (at least above 0.5 kHz).
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump