normalize to maximum volume

Discussion forum for Amadeus users

Moderator: Martin Hairer

Post Reply
jiggy1965
Posts: 6
Joined: Fri Jan 31, 2014 12:04 pm

normalize to maximum volume

Post by jiggy1965 »

I've got a track which is a bit soft, so I want to normalize it so the track plays in its maximum volume without peaking. I've got a problem understanding the normalize settings. But you know what I mean: the entire track should go up in volume evenly but the largest peak in the waveform shouldn't go over the maximum value. In the end the waveform shouldn't end up at it's maximum volume.

I've done a waveform analysis. These are the values:
Image

When I go to Normalize I don't quite understand when to use dB or % or sample value or RMS value.
Image

But based on the waveform analysis, how do I get to the db level value in the normalize window and which reference value should I pick?

So that my track is up-volumed nice and evenly and to its maximum while staying in the green?

User avatar
Jim Edgar
Posts: 118
Joined: Mon Mar 01, 2010 7:20 am
Contact:

normalize to maximum volume

Post by Jim Edgar »

The best way to Normalize without peaking is to choose the peak value option (Maximal Sample Value).  db or % are just two scales.  -0db is the peak, or 100%.


I've never had good results using RMS - but basically you are setting the point for the average energy, without regard to clipping.



If I'm reading your results correctly, you are already peaking (going above -0 db) which is why you have clipped samples.



Usually, I will Normalize to -3 db, which leaves enough headroom not to cause any issues on different playback systems. 


If you have significant fluctuations in recording levels (i.e. a large dynamic range - which it appears you do since minimum is -infinity), then you might want to use Compression (Dynamics Processor) first.  This will bump up the lows and bring down the highs.  Then you can Normalize to a specific level.  For my purposes - VO/Narration - this is a much more useful way of working. Note - the compression step will also tend to raise the noise floor in a recording.


Not sure if this is your raw recording file, but if so, you should record so that your peaks are lower - between -6 db and -3 db is a good place to aim.  The clipped samples are distorted and that is very difficult to undo.


Hope that helps,


- Jim



jimedgarvoices.com / @jimedgarvoices

Sarah k Alawami

normalize to maximum volume

Post by Sarah k Alawami »

I always use db. You can never get above 0 or you will distort at the loudest part. Think of normalization as turning up a volume knob. the loudest part goes to 0, the rest goes up in proportion to that. This is how I was taught ibn the courses I took at a community college.

Hope that helps a bit.
On Feb 2, 2014, at 2:43 AM, jiggy1965 <forum2mail@hairersoft.com> wrote:
I've got a track which is a bit soft, so I want to normalize it so the track plays in its maximum volume without peaking. I've got a problem understanding the normalize settings. But you know what I mean: the entire track should go up in volume evenly but the largest peak in the waveform shouldn't go over the maximum value. In the end the waveform shouldn't end up at it's maximum volume.

I've done a waveform analysis. These are the values:


When I go to Normalize I don't quite understand when to use dB or % or sample value or RMS value.


But based on the waveform analysis, how do I get to the db level value in the normalize window and which reference value should I pick?

So that my track is up-volumed nice and evenly and to its maximum while staying in the green?




_______________________________________________
Amadeus forum mailing list
Unsubscribe / change settings at http://two.pairlist.net/mailman/listinfo/forum_list
_______________________________________________
Amadeus forum mailing list
Unsubscribe / change settings at http://two.pairlist.net/mailman/listinfo/forum_list

jiggy1965
Posts: 6
Joined: Fri Jan 31, 2014 12:04 pm

Post by jiggy1965 »

My track starts of a bit soft and louder in the middle. I normalized at a -0.1 db setting and used 'sample rate' in stead of rms (which never seems to work for me as it often makes the tracks softer in volume). Which already did a decent job of making it a bit louder.

Still starts out a bit soft though. I would like the track to be just as loud the entire track. You mentioned the Dynamics Processor would do that? So I would first have to use Dynamics Processor at some preset setting (which?) to make the volume about the same all over and then use Normalize to make it close to 0 db in proportion the entire track?

User avatar
Jim Edgar
Posts: 118
Joined: Mon Mar 01, 2010 7:20 am
Contact:

normalize to maximum volume

Post by Jim Edgar »

Yes, if you want to smooth out a wide dynamic range, you should use the Dynamics Processor before you Normalize (though you might Normalize to begin, then use the Dynamics Processor, then Normalize to Master Level - depends upon what you are doing).


A good general Dynamics Processor setting to get you in the ballpark would be Threshold: -8 - -4 db, Ratio: 1.5 - 3.0 : 1, Attack <1 ms, Decay ~40 ms


That would be enough to smooth things out without being too noticeable.  But, you should tune to your needs. 

And of course, work on a copy of the file while you are dialing in the settings to your taste. 


-0.1 db is very close to distorting and leaves you little room to go anywhere. You may find you don't need to Normalize that hot once you tame the Dynamics.  It's not uncommon for Compression (Dynamics Processor) to reduce the track's overall volume, which is why you want to Normalize (to the peak, not the average RMS) afterwards.


Hope that helps!


- Jim


jimedgarvoices.com / @jimedgarvoices

jiggy1965
Posts: 6
Joined: Fri Jan 31, 2014 12:04 pm

Post by jiggy1965 »

I'm new to this effects settings like dynamics processor. Can't find all the settings you're describing.

In the graphic there are four circles I can drag around. All describe a Threshold setting. The 2 on the left also mention Ratio and the 2 on the left also mention Headroom.

So there are 4 bullets mentioning Threshold, 2 also Ratio and 2 also Headroom. So what would be their acceptable values? :-)

I can see Attack below with its slider. Decay is the same as Release Time? (as this is the only other value in ms).

Image

User avatar
Jim Edgar
Posts: 118
Joined: Mon Mar 01, 2010 7:20 am
Contact:

normalize to maximum volume

Post by Jim Edgar »

Sorry - was describing the Compression settings from memory, which are more for hardware controls.

The AU Dynamics Processor should give you a window similar to the attached image - these settings are not precisely what I said, but will give you a place to start from.


hope that helps.
- Jim

jiggy1965
Posts: 6
Joined: Fri Jan 31, 2014 12:04 pm

Post by jiggy1965 »

The problem is that with most of my mp3's I'm never quite able to normalize them so that the maximum volume is used.

For example, I'm trying it with this random track of Duck Ellington:
http://sharesend.com/rqu9jprm

When opened you'll the the waveform. Though analysis says the maximum peak is around -0,5 db, the waveform doesn't quite seem to read that. If fact, while playing and looking at the playback controller, it's loudest moments reach about -3 db. Only if I use the Find Maximum option I can see the waveform reach about -1 db.

But anyway, I want the entire track to have that average volume of about -1db. So first I use the Dynamic Processor effect with a preset of Hard to equal things out. That seems to straighten things out, but decreases the volume considerably. The peak maximum becomes -9,5 db, the average RMS power -23 db.

But the overall volume seems to look equal. So now I use Normalize. To a setting of -1db of the maximum sample value. The volume increases, but now seems to be lower in volume than the original was. The peak maximum now reaches -1db (so far so good), but the waveform now looks more like the maximum is about -6 db. Which seems about right when I play it and look at the playback controller which seems to go not higher than about -6 db.

What I want to do is give my track from start to finish about the same volume and then normalize it so that the volume becomes on average about -1 db and not peaking above that.

The only files I seem to get this to work is like mp3s of perfect music notes like a mp3 of a A-note which has a perfect sine waveform. But with all my music mp3s, using the dynamic processor to straighten things out and then normalizing them only seems to make them lower in volume.

What am I doing wrong and how can I achieve what I want to do? Can you try it yourself with above music file perhaps?

CDJonah
Posts: 119
Joined: Sat Nov 11, 2006 12:10 am

normalize to maximum volume

Post by CDJonah »

Certainly the -.4 looks pretty reasonable to me. If you zoom way in at
the maximum value, you see spikes, each about 1 channel wide about 20 ms
apart. With those sorts of spikes there is no way that you can do what
you want without removing the narrow high-frequency swings.

Chuck

On 2/6/14 4:53 PM, jiggy1965 wrote:
The problem is that with most of my mp3's I'm never quite able to normalize them so that the maximum volume is used.

For example, I'm trying it with this random track of Duck Ellington:
http://sharesend.com/rqu9jprm

When opened you'll the the waveform. Though analysis says the maximum peak is around -0,5 db, the waveform doesn't quite seem to read that. If fact, while playing and looking at the playback controller, it's loudest moments reach about -3 db. Only if I use the Find Maximum option I can see the waveform reach about -1 db.

But anyway, I want the entire track to have that average volume of about -1db. So first I use the Dynamic Processor effect with a preset of Hard to equal things out. That seems to straighten things out, but decreases the volume considerably. The peak maximum becomes -9,5 db, the average RMS power -23 db.

But the overall volume seems to look equal. So now I use Normalize. To a setting of -1db of the maximum sample value. The volume increases, but now seems to be lower in volume than the original was. The peak maximum now reaches -1db (so far so good), but the waveform now looks more like the maximum is about -6 db. Which seems about right when I play it and look at the playback controller which seems to go not higher than about -6 db.

What I want to do is give my track from start to finish about the same volume and then normalize it so that the volume becomes on average about -1 db and not peaking above that.

The only files I seem to get this to work is like mp3s of perfect music notes like a mp3 of a A-note which has a perfect sine waveform. But with all my music mp3s, using the dynamic processor to straighten things out and then normalizing them only seems to make them lower in volume.

What am I doing wrong and how can I achieve what I want to do? Can you try it yourself with above music file perhaps?




_______________________________________________
Amadeus forum mailing list
Unsubscribe / change settings at http://two.pairlist.net/mailman/listinfo/forum_list

_______________________________________________
Amadeus forum mailing list
Unsubscribe / change settings at http://two.pairlist.net/mailman/listinfo/forum_list

jiggy1965
Posts: 6
Joined: Fri Jan 31, 2014 12:04 pm

Post by jiggy1965 »

Yes, and it's that kind of extreme spikes that make normalizing so difficult. I've opened a few mp3's and the waveforms move around the -4db for example. But they also contain spikes that come close or even peak over 0 db. So when I normalize them to make the overall volume as close to 0 db as possible, those spikes prevent them from getting much louder than -4 db.

So when I listen to a track that is on average close to e.g. -1 db and then a track that is on average -4 db I always have to turn up the volume.

Tried to lower those spikes by using Dynamic Processor after Normalize, but that makes it less loud again. Also Normalized first and then used the Dynamic Processor, but that made it less loud also.

What I really would want to so is to get the overall volume the same. About -1 db for example. So that all my tracks are just as loud and have the maximum volume. And then to decrease the spikes that are peaking above 0 db, to bring those back to about -1 db also.

Could I do that some how using one of the effects?

CDJonah
Posts: 119
Joined: Sat Nov 11, 2006 12:10 am

normalize to maximum volume

Post by CDJonah »

As no matter what you do at this point will affect the sound. You could
try either the 10-band equalizer or the graphic equalizer as part of the
au units (cut the high frequencies). You could try just try overdriving
the thing and clipping. It is up to you how badly you want to distort
the sound.

Chuck
On 2/7/14 3:44 AM, jiggy1965 wrote:
Yes, and it's that kind of extreme spikes that make normalizing so difficult. I've opened a few mp3's and the waveforms move around the -4db for example. But they also contain spikes that come close or even peak over 0 db. So when I normalize them to make the overall volume as close to 0 db as possible, those spikes prevent them from getting much louder than -4 db.

So when I listen to a track that is on average close to e.g. -1 db and then a track that is on average -4 db I always have to turn up the volume.

Tried to lower those spikes by using Dynamic Processor after Normalize, but that makes it less loud again. Also Normalized first and then used the Dynamic Processor, but that made it less loud also.

What I really would want to so is to get the overall volume the same. About -1 db for example. So that all my tracks are just as loud and have the maximum volume. And then to decrease the spikes that are peaking above 0 db, to bring those back to about -1 db also.

Could I do that some how using one of the effects?




_______________________________________________
Amadeus forum mailing list
Unsubscribe / change settings at http://two.pairlist.net/mailman/listinfo/forum_list

_______________________________________________
Amadeus forum mailing list
Unsubscribe / change settings at http://two.pairlist.net/mailman/listinfo/forum_list

User avatar
Jim Edgar
Posts: 118
Joined: Mon Mar 01, 2010 7:20 am
Contact:

normalize to maximum volume

Post by Jim Edgar »

Tried to lower those spikes by using Dynamic Processor after Normalize, but that makes it less loud again. Also Normalized first and then used the Dynamic Processor, but that made it less loud also.


No. You need to use the Dynamic Processor to compress the range (i.e. lower the highs and expand the lows).  Then Normalize after that. 


Between those two steps in that order, you should end up with a track which has a smaller dynamic range (difference between soft and loud bits) and a higher overall volume. 


And I really would not recommend clipping the signal.  "Overdriving" and digital do not play well together.



hope that gets you back on track.


- J

 

Post Reply