Become a Site Supporter and Never see Ads again!

Author Topic: aplying gain vs. normalizing  (Read 13737 times)

0 Members and 1 Guest are viewing this topic.

Offline SparkE!

  • Trade Count: (0)
  • Taperssection Member
  • ***
  • Posts: 773
Re: aplying gain vs. normalizing
« Reply #15 on: June 12, 2006, 01:55:42 PM »
Teddy, I think you've got it backwards.  Normalizing does not mess with the dynamic range at all.  As you said, It just amplifies everything by the same amount with the result that the largest sample goes to -0 dB.  There is still the same amount of difference (dB-wise) between the quiet sections and the loud sections.  When you use RMS normalization, you compress your original wide dynamic range so that the orignally quiet sections are closer in volume to the loud sections.  That is by definition a reduction in dynamic range.
How'm I supposed to read your lips when you're talkin' out your ass? - Lern Tilton

Ignorance in audio is exceeded only by our collective willingness to embrace and foster it. -  Srajan Ebaen

Offline greenone

  • Trade Count: (4)
  • Needs to get out more...
  • *****
  • Posts: 9273
  • Gender: Male
  • Russian mics... strong like bull...
Re: aplying gain vs. normalizing
« Reply #16 on: June 12, 2006, 02:07:50 PM »
I've got a question I've never seen answered elsewhere...why do I see so many info files that say "normalized to 98%"? Why wouldn't you want to go to 100%? I mean, isn't 100% by definition not distorted?
Unofficial Blues Traveler archivist - glad to work on any BT or related recordings
archive.org admin - happy to upload tracked material to the LMA

Offline admkrk

  • Trade Count: (0)
  • Taperssection All-Star
  • ****
  • Posts: 1733
  • I'm an idiot
Re: aplying gain vs. normalizing
« Reply #17 on: June 12, 2006, 02:31:43 PM »
Teddy, I think you've got it backwards.  Normalizing does not mess with the dynamic range at all.  As you said, It just amplifies everything by the same amount with the result that the largest sample goes to -0 dB.  There is still the same amount of difference (dB-wise) between the quiet sections and the loud sections.  When you use RMS normalization, you compress your original wide dynamic range so that the orignally quiet sections are closer in volume to the loud sections.  That is by definition a reduction in dynamic range.

and now i'm even more confused.  i remember seeing a warning, in sf i think, about how normlizing w/ rms could introduce clipping. to me that sounds more like the way teddy said it. also might answer greenone's question?
"the faster you go ahead, the behinder you get"

"If you can drink ram's piss, fuck, you can drink anything"

Offline Scooter

  • Trade Count: (2)
  • Taperssection All-Star
  • ****
  • Posts: 1770
  • Gender: Male
Re: aplying gain vs. normalizing
« Reply #18 on: June 12, 2006, 03:18:07 PM »
Teddy, I think you've got it backwards.  Normalizing does not mess with the dynamic range at all.  As you said, It just amplifies everything by the same amount with the result that the largest sample goes to -0 dB.  There is still the same amount of difference (dB-wise) between the quiet sections and the loud sections.  When you use RMS normalization, you compress your original wide dynamic range so that the orignally quiet sections are closer in volume to the loud sections.  That is by definition a reduction in dynamic range.

QFT.  you nailed it.
MBHO 603a(ka200n/ka500hn) >
R-44, or H120

LMA Recordings

RebelRebel

  • Guest
  • Trade Count: (0)
Re: aplying gain vs. normalizing
« Reply #19 on: June 12, 2006, 03:37:31 PM »
yes, I had it backwards. My mistake. wisdom teeth extraction/vicodin and interweb dont mix.


Teddy, I think you've got it backwards.  Normalizing does not mess with the dynamic range at all.  As you said, It just amplifies everything by the same amount with the result that the largest sample goes to -0 dB.  There is still the same amount of difference (dB-wise) between the quiet sections and the loud sections.  When you use RMS normalization, you compress your original wide dynamic range so that the orignally quiet sections are closer in volume to the loud sections.  That is by definition a reduction in dynamic range.

QFT.  you nailed it.

Offline SparkE!

  • Trade Count: (0)
  • Taperssection Member
  • ***
  • Posts: 773
Re: aplying gain vs. normalizing
« Reply #20 on: June 12, 2006, 03:56:11 PM »

wisdom teeth extraction/vicodin and interweb dont mix.


Ouch.  Hopefully the extraction went well?  I've known people whose cheeks have swelled up like a chipmunk's from an extraction gone bad.
How'm I supposed to read your lips when you're talkin' out your ass? - Lern Tilton

Ignorance in audio is exceeded only by our collective willingness to embrace and foster it. -  Srajan Ebaen

Offline morningdew

  • Trade Count: (0)
  • Taperssection Member
  • ***
  • Posts: 252
Re: aplying gain vs. normalizing
« Reply #21 on: June 13, 2006, 08:28:18 AM »
Ok, so this thread won't go down as a great for the archive. :o.

Here is what I've read so far:

Quote
Metanormalize(wavelab tool) normalizes using RMS(average levels)
use the Meta Normalize tool for the best results(use instead of change gain)

Quote
Adding gain is much better, IMHO. The dynamic levels are preserved, and it is really simple to do...with normalization, you can possibly screw up the dynamics.

Quote
RMS normalization can screw up the dynamics.  adding gain(straight peak normalization) won't,

So, have we come to conclusion or is the conclusion no one really knows?

Here is the scenario.  I make a recording.  My levels are low by let's say 5 dB.  I didn't touch or mess with the levels during my whole recording so I just want to highlight the whole file and raise the levels to 0 dB or -0.2 dB (I've read a lot that people do this, why I don't know and this was mentioned in a previous post, i.e. normalize to 98%).  In SoundForge I have 2 choices.

1. Normalize to my choice of dB.
2. RMS normalization.

Which is the best choice in this situation?

Offline Scooter

  • Trade Count: (2)
  • Taperssection All-Star
  • ****
  • Posts: 1770
  • Gender: Male
Re: aplying gain vs. normalizing
« Reply #22 on: June 13, 2006, 09:13:15 AM »
MBHO 603a(ka200n/ka500hn) >
R-44, or H120

LMA Recordings

Offline SparkE!

  • Trade Count: (0)
  • Taperssection Member
  • ***
  • Posts: 773
Re: aplying gain vs. normalizing
« Reply #23 on: June 13, 2006, 09:16:15 AM »
I just want to highlight the whole file and raise the levels to 0 dB or -0.2 dB (I've read a lot that people do this, why I don't know and this was mentioned in a previous post, i.e. normalize to 98%).  In SoundForge I have 2 choices.

1. Normalize to my choice of dB.
2. RMS normalization.

Which is the best choice in this situation?


In my opinion, the best choice is to normalize to -0 dB.  By -0, I mean as close to 0.0 dB as possible without going over.  Dithering can cause you to go over, so some people don't like to normalize to absolute full scale.  On the other hand, I'm not sure that you'd notice if you clipped the dithering signal.  After all, it's supposedly inaudible noise.

If you use RMS normalization, you run the risk of actually clipping the audio signal itself.  I think some RMS routines employ soft clipping in order to keep the waveform between the rails instead of allowing them to hard clip, but I can't say that for certain.  I'm relying on inferences from things that other people have said in discussions similar to this one.  In particular, I don't know how SoundForge handles the clipping aspect of normalization.
How'm I supposed to read your lips when you're talkin' out your ass? - Lern Tilton

Ignorance in audio is exceeded only by our collective willingness to embrace and foster it. -  Srajan Ebaen

Offline scb

  • Eli Manning should die of gonorrhea and rot in hell. Would you like a cookie, son?
  • Trade Count: (11)
  • Needs to get out more...
  • *****
  • Posts: 8676
  • Gender: Male
Re: aplying gain vs. normalizing
« Reply #24 on: June 13, 2006, 10:10:18 AM »
So, have we come to conclusion or is the conclusion no one really knows?

RMS raises the levels based on the RMS level.  Everything above the RMS level is either clipped or compressed, depending on how the software wants to do it (which is why i don't recommend RMS normalization)

Offline morningdew

  • Trade Count: (0)
  • Taperssection Member
  • ***
  • Posts: 252
Re: aplying gain vs. normalizing
« Reply #25 on: June 13, 2006, 10:39:40 AM »
Ok, thanks guys.

I believe we have a good conclusion that in this case it is best to use "plain" normalization. Aviod using RMS normalization.

Thanks.

Offline admkrk

  • Trade Count: (0)
  • Taperssection All-Star
  • ****
  • Posts: 1733
  • I'm an idiot
Re: aplying gain vs. normalizing
« Reply #26 on: June 13, 2006, 10:45:43 AM »
Ok, thanks guys.

I believe we have a good conclusion that in this case it is best to use "plain" normalization. Aviod using RMS normalization.

Thanks.


that's what i got out of it.
"the faster you go ahead, the behinder you get"

"If you can drink ram's piss, fuck, you can drink anything"

Offline Church-Audio

  • Trade Count: (44)
  • Needs to get out more...
  • *****
  • Posts: 7571
  • Gender: Male
Re: aplying gain vs. normalizing
« Reply #27 on: June 14, 2006, 01:24:51 AM »
Normalizing takes the average sound level and brings it up, it does this by looking at the peeks and looking at the valleys and using RMS root mean square or other algorithm, it comes up with a maximum level in can bring the recording up to before distortion. Gain is absolute, it is blind it will not ask you to scan a section for an RMS sample; it will simply apply gain to the whole section. It can bring up noise floor if used improperly; it can also bring up distortion levels too.

In order to use GAIN you must look at the loudest peek your self and decide how much gain can be added to it, before it distorts. The best way for a beginner to get more gain IMO is to normalize.

You must take songs and do them one at a time. I even go as far as to normalize quiet sections and loud sections separately, and then I normalize the whole song again just to smooth out the transition. I always keep a safety copy of the file as a back up.

 In the good old days of tape there was no such thing as a back up. Now you can play with EQ, Normalize, Expand and Compress till your hearts content.

 I know this will not help you much but the best advice is to take a really bad recording and see how much you can improve it. And one more tip try listening to this "improved" track with as many different speakers, Headphones as possible and try and get it so it sounds pretty good on everything and I bet you will have a nice recording when you’re done. Good luck

And remember there are no mistakes in audio just fun adventures :)



Chris Church

well thanks for ya'lls help, i think i might be even more comfused now  ???

i went the rms route and all seems fine. for what ever reason i couldn't change the rt channel independently, only the left. at least the peaks were w/ in 1db so i can live w/ it.
for warranty returns email me at
EMAIL Sales@church-audio.com

Offline poorlyconditioned

  • Trade Count: (0)
  • Taperssection All-Star
  • ****
  • Posts: 1958
  • I'm a tapir!
Re: aplying gain vs. normalizing
« Reply #28 on: June 14, 2006, 01:31:38 AM »
Normalizing takes the average sound level and brings it up, it does this by looking at the peeks and looking at the valleys and using RMS root mean square or other algorithm, it comes up with a maximum level in can bring the recording up to before distortion. Gain is absolute, it is blind it will not ask you to scan a section for an RMS sample; it will simply apply gain to the whole section. It can bring up noise floor if used improperly; it can also bring up distortion levels too.

In order to use GAIN you must look at the loudest peek your self and decide how much gain can be added to it, before it distorts. The best way for a beginner to get more gain IMO is to normalize.

You must take songs and do them one at a time. I even go as far as to normalize quiet sections and loud sections separately, and then I normalize the whole song again just to smooth out the transition. I always keep a safety copy of the file as a back up.

 In the good old days of tape there was no such thing as a back up. Now you can play with EQ, Normalize, Expand and Compress till your hearts content.

 I know this will not help you much but the best advice is to take a really bad recording and see how much you can improve it. And one more tip try listening to this "improved" track with as many different speakers, Headphones as possible and try and get it so it sounds pretty good on everything and I bet you will have a nice recording when you’re done. Good luck

And remember there are no mistakes in audio just fun adventures :)



Chris Church

well thanks for ya'lls help, i think i might be even more comfused now  ???

i went the rms route and all seems fine. for what ever reason i couldn't change the rt channel independently, only the left. at least the peaks were w/ in 1db so i can live w/ it.

Hey Chris.

I'm still confused!  Does the RMS method apply a uniform gain over the whole sample or does it vary it over time?  OK, if it varies the gain, do you know what algorithm is used?

Thanks,
   Richard
Mics: Sennheiser MKE2002 (dummy head), Studio Projects C4, AT825 (unmodded), AT822 franken mic (x2), AT853(hc,c,sc,o), Senn. MKE2, Senn MKE40, Shure MX183/5, CA Cards, homebrew Panasonic and Transsound capsules.
Pre/ADC: Presonus Firepod & Firebox, DMIC20(x2), UA5(poorly-modded, AD8620+AD8512opamps), VX440
Recorders: Edirol R4, R09, IBM X24 laptop, NJB3(x2), HiMD(x2), MD(1).
** This individual has moved to user "illconditioned" **

Offline Scooter

  • Trade Count: (2)
  • Taperssection All-Star
  • ****
  • Posts: 1770
  • Gender: Male
Re: aplying gain vs. normalizing
« Reply #29 on: June 14, 2006, 09:36:51 AM »
somebody correct me if i'm wrong, but i believe RMS mode takes the root mean square of a selected portion of audio over time, and normalizes it using a compression so that it has an everage level that you select.  i don't know what the attack,release settings would be fot that though...
« Last Edit: June 14, 2006, 09:39:08 AM by Scooter »
MBHO 603a(ka200n/ka500hn) >
R-44, or H120

LMA Recordings

 

RSS | Mobile
Page created in 0.093 seconds with 44 queries.
© 2002-2024 Taperssection.com
Powered by SMF