Quantcast
ETC - Frequency Range - Gearspace.com
The No.1 Website for Pro Audio
ETC - Frequency Range
Old 23rd January 2013
  #1
Lives for gear
 
jim1961's Avatar
 
🎧 5 years
ETC - Frequency Range

ETC - Frequency Range-01-23-etc-450-5.6k-vs-full-bandwidth.gif

My question is, what frequency range should be used? Do you use?

The above graphs shows in black the full bandwidth while the blue shows 450hz - 5.6k. I used 200us smoothing for comparative purposes. Unsmoothed ETC's show the same sort of spread, about 5db in most places.

Whats disturbing to me the how much the magnitude changes. Is this suppose to work this way? Does it change like this in other software?
Old 23rd January 2013
  #2
Lives for gear
 
akebrake's Avatar
 
🎧 5 years
White spectrum

Quote:
Originally Posted by jim1961 ➡️

My question is, what frequency range should be used? Do you use?

The above graphs shows in black the full bandwidth while the blue shows 450hz - 5.6k. I used 200us smoothing for comparative purposes. Unsmoothed ETC's show the same sort of spread, about 5db in most places.

Whats disturbing to me the how much the magnitude changes. Is this suppose to work this way? Does it change like this in other software?
Good questions and I wondered exactly the same untill I red Tooles book and also asked a number of AES & Syn Aud Con guys about it.

You cannot estimate level from ETC unless same spectrum of e.g.direct and reflection.

I’ll take the oportunity to present my view (understanding?) just to check if I got it reasonable right.
(I am not a math guy or programmer)

An Envelope Time Curve (ETC) has a white spectrum weighting to it.
E.g. equal energy per Hz. (It "favours" hi frequncies) On the ETC you see a lot of treble and the magnitude is falling towards low frequencies.

If you post process the impulse with a pink filter first (-3dB/oct slope) you can then filter it again with 1/1 octave or 1/3 octave etcetera.
This is done inside the program.

The attached diagram (WinMLS) is an overlay of three bandpass filtered ETC:s without normalizing. Note that mid and lo frequencies are ”later” and lower in magnitude.(Spread out) Small Genelec 1029 at 0.75 m distance. Ceiling and floor reflections together after Direct.

(the lowest green curve is difficult to see. Ill try to find a better image but you may be get my point anyway)

Note: The magnitude of individual frequency bands are still uncertain because we are in the timedomain and lower frequencies are less defined in time.

Layman descripion. Your turn…

May be JohnPM or Chris Liscio can comment

Chers

ake
Attached Images
File Type: jpg 1029 on Radio Table Metzy.jpg (99.5 KB, 520 views)

Last edited by akebrake; 21st February 2013 at 11:07 PM.. Reason: wrong colour and I am not shouting
Old 23rd January 2013
  #3
Lives for gear
 
🎧 10 years
SAC discussed this very aspect some time ago here; i'll try and dig the bookmark this evening.
Old 23rd January 2013 | Show parent
  #4
Lives for gear
 
jim1961's Avatar
 
🎧 5 years
Quote:
Originally Posted by akebrake ➡️
Good questions and I wondered exactly the same untill I red Tooles book and also asked a number of AES & Syn Aud Con guys about it.

You cannot estimate level from ETC unless same spectrum of e.g.direct and reflection.

I’ll take the oportunity to present my view (understanding?) just to check if I got it reasonable right.
(I am not a math guy or programmer)

An Envelope Time Curve (ETC) has a white spectrum weightto it.
E.g. equal energy per Hz. (It "favours" hi frequncies) On the ETC you see a lot of treble and the magnitude is falling towards low frequencies.

If you post process the impulse with a pink filter first (-3dB/oct slope) you can then filter it again with 1/1 octave or 1/3 octave etcetera.
This is done inside the program.

The attached diagram (WinMLS) is an overlay of three bandpass filtered ETC:s without normalizing. Note that mid and lo frequencies are ”later” and lower in magnitude.(Spread out) Small Genelec 1029 at 0.75 m distance. Ceiling and floor reflections together after Direct.

(the lowest black curve is difficult to see. Ill try to find a better image but you may be get my point anyway)

Note: The magnitude of individual frequency bands are still uncertain because we are in the timedomain and lower frequencies are less defined in time.

Layman descripion. Your turn…

May be JohnPM or Chris Liscio can comment

Chers

ake
I bolded a part of your response that I didnt know or consider before. Interesting. So is that to say a 450hz-5.6k sliced ETC implies an incorrect amplitude?

Is that to also say one octave bands may be accurate relative to each other, but their absolute magnitude cant be derived from such a slice?

And, that only a full bandwidth ETC shows the correct magnitude relative to the direct signal?

In my case, I am trying to determine the REAL level(s) of my ISD gap. The finer I slice it, the higher the returns tend to be. But slicing is seemingly the only way to get an idea of the frequency response within the time/magnitude domain. The only way I know how to see how even the reflections are at different frequencies.
Old 23rd January 2013
  #5
Lives for gear
 
akebrake's Avatar
 
🎧 5 years
Jim,
I'am comming back on this one to morrow and I will check some earlier posts as well.

Best
Old 24th January 2013
  #6
Gear Guru
 
1 Review written
🎧 15 years
Direction

Quote:
In my case, I am trying to determine the REAL level(s) of my ISD gap. The finer I slice it, the higher the returns tend to be. But slicing is seemingly the only way to get an idea of the frequency response within the time/magnitude domain. The only way I know how to see how even the reflections are at different frequencies.
Real as perceived by a microphone or the ear, or two ears?

I don't have references readily available but I would expect our hearing to be a bit to deaf to HF arriving from behind. Gradually less so as the source becomes lateral.

DD
Old 24th January 2013 | Show parent
  #7
Lives for gear
 
jim1961's Avatar
 
🎧 5 years
Quote:
Originally Posted by DanDan ➡️
Real as perceived by a microphone or the ear, or two ears?

I don't have references readily available but I would expect our hearing to be a bit to deaf to HF arriving from behind. Gradually less so as the source becomes lateral.

DD
I suppose I lacked clarity in that statement.

My bumps around 14-18ms are at 150 degrees, the ones around 10ms are at 100 degrees, and the ones around 6ms are at 80 degrees to address your directionality thought.

I want my ISD gap to be -25db or better. I will live with the floor bounce being at -20db at 2ms. The clutter around 1ms I think is cabinet related, and will be addressed at a later point.

Basically, using sliced ETC's, I am not quite there. Using the Fullrange ETC, I am. Thus, the thread question.

ETC - Frequency Range-01-24-etc-450-5.6k-right.gif

ETC - Frequency Range-01-24-etc-fullrange-right.gif

The blue plot is my 450hz-5.6k slice. The black one is the full range ETC.
Old 25th January 2013 | Show parent
  #8
Lives for gear
 
akebrake's Avatar
 
🎧 5 years
Quote:
Originally Posted by localhost127 ➡️
SAC discussed this very aspect some time ago here; i'll try and dig the bookmark this evening.
Looking forwrd to it!

ake
Old 25th January 2013 | Show parent
  #9
Lives for gear
 
akebrake's Avatar
 
🎧 5 years
Quote:
Originally Posted by jim1961 ➡️
I suppose I lacked clarity in that statement.

My bumps around 14-18ms are at 150 degrees, the ones around 10ms are at 100 degrees, and the ones around 6ms are at 80 degrees to address your directionality thought.

I want my ISD gap to be -25db or better. I will live with the floor bounce being at -20db at 2ms. The clutter around 1ms I think is cabinet related, and will be addressed at a later point.

Basically, using sliced ETC's, I am not quite there. Using the Fullrange ETC, I am. Thus, the thread question.

Attachment 327015

Attachment 327016

The blue plot is my 450hz-5.6k slice. The black one is the full range ETC.
Jim,

I think you have partly answered your question yourself.

With the wider range ETC you put more energy (number of Hz) into the measurement (remember the display shows a white spectrum) and your gap looks lower in magnitude.

AFAIK a band restricted measurement will show more of the low frequencies and a lower level compared to a fullrange sweep without normalizing.
This ”band restriction” is usually performed in different ways in postprocessing like 1/1octave, 1/3:rd octave or other filtering depending of software functions.
From 0 to 100 Hz there is only 100 Hz. From 100-1000 Hz there is 900 Hz and from 1000- 10kHz we have 9000 Hz
(a lot more energy with constant BW filtering eg. 10Hz)

More light shed?

Best
Old 25th January 2013 | Show parent
  #10
Lives for gear
 
jim1961's Avatar
 
🎧 5 years
Quote:
Originally Posted by akebrake ➡️
Jim,

I think you have partly answered your question yourself.

With the wider range ETC you put more energy (number of Hz) into the measurement (remember the display shows a white spectrum) and your gap looks lower in magnitude.

AFAIK a band restricted measurement will show more of the low frequencies and a lower level compared to a fullrange sweep without normalizing.
This ”band restriction” is usually performed in different ways in postprocessing like 1/1octave, 1/3:rd octave or other filtering depending of software functions.
From 0 to 100 Hz there is only 100 Hz. From 100-1000 Hz there is 900 Hz and from 1000- 10kHz we have 9000 Hz
(a lot more energy with constant BW filtering eg. 10Hz)

More light shed?

Best
Yes. I think I get it.

So in the end, the full bandwidth ETC shows the proper amplitude of the reflections due to it matching the bandwidth of the direct signal. Yes?
Old 25th January 2013 | Show parent
  #11
Lives for gear
 
jim1961's Avatar
 
🎧 5 years
Quote:
Originally Posted by localhost127 ➡️
SAC discussed this very aspect some time ago here; i'll try and dig the bookmark this evening.
That would be a good read
Old 25th January 2013 | Show parent
  #12
Lives for gear
 
akebrake's Avatar
 
🎧 5 years
Quote:
Originally Posted by jim1961 ➡️
Yes. I think I get it.

So in the end, the full bandwidth ETC shows the proper amplitude of the reflections due to it matching the bandwidth of the direct signal. Yes?
No unfortunately not. (bolded mine) If Matching spectrum, yes eventually a little better but how to achive that?

Floyd Toole wrote: (Sound Reproduction) p 257
Quote:
”In measurements of reflections we need to measure the spectrum level of reflections to be able to gauge their relative audible effect.” <Snip>
” Whenever the direct and reflected sounds have different spectra the simple broadband ETC or impulseresponses are not thrustworthy indicators of audible effects”.
I don’t think the ”reflection” (e.g. a number of reflections in the ISD-Gap have the same spectrum as the loudspeaker on axis. Side walls, floor and ceiling are off axis to the speaker and also treated with different absorbers.
The microphone is probably not ”pointed” equally at direct and reflecting surfaces.

We need "frequency response" at the time in question to be able to measure "proper amplitude"

Best
Old 25th January 2013 | Show parent
  #13
Lives for gear
 
jim1961's Avatar
 
🎧 5 years
Quote:
Originally Posted by akebrake ➡️
No unfortunately not. (bolded mine) If Matching spectrum, yes eventually a little better but how to achive that?

Floyd Toole wrote: (Sound Reproduction) p 257

I don’t think the ”reflection” (e.g. a number of reflections in the ISD-Gap have the same spectrum as the loudspeaker on axis. Side walls, floor and ceiling are off axis to the speaker and also treated with different absorbers.
The microphone is probably not ”pointed” equally at direct and reflecting surfaces.

We need "frequency response" at the time in question to be able to measure "proper amplitude"

Best
Perhaps there is a greater context to which the quoted Toole statement was made, but it seems he is basically saying that the ETC is unreliable given nearly any reflected response is going to differ in spectra compared to the direct response.

But as to what most people use the ETC for, that is the measured amplitude of reflections in a time domain, the question persists as to what ETC parameters (full bandwidth / slices) we use to comply with the LEDE / RFZ criteria for an ISD gap amplitude using the ETC. If -20db/-25db/-30db is what is sought, then is a full bandwidth ETC, despite its flaws, the proper ETC constraint (full bandwidth) to gauge whether the criteria (ISD gap amplitude) is met or not?
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 19 views: 8642
Avatar for SnakeCained
SnakeCained 14th January 2006
replies: 4 views: 4545
Avatar for spicemix
spicemix 20th July 2010
Post Reply

Welcome to the Gearspace 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…

Forum Jump
Forum Jump