Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > General > Audio encoding

Reply
 
Thread Tools Search this Thread Display Modes
Old 7th May 2018, 19:02   #161  |  Link
LoRd_MuldeR
Software Developer
 
LoRd_MuldeR's Avatar
 
Join Date: Jun 2005
Location: Last House on Slunk Street
Posts: 13,032
Quote:
Originally Posted by kolamorx View Post
Hello.
I've read your documentation about the maximum gain factor, that ususally it's not recommended to increase the value above the default but sometimes it may be necessary to allow higher gain factors.
So if I understood correctly (and from my experience), when you have to allow higher gain factors, it's better to set it as less value as possible.
No matter what you do, the gain factor for each frame will never be allowed to exceed the frame's local max. gain factor – the maximum that this particular frame could be amplified without any clipping (distortion).

Furthermore, the actual gain factor applied to a frame will usually be even lower than the "local max. gain factor", because we apply a minimum filter followed by a Gaussian smoothing filter to each "neighborhood" (sliding window) of frames.

The "max. gain factor" option is an additional hard limit on the maximum gain factor that can be applied to frame. This mostly exists to avoid humongous amplification factors in extremely quiet (almost silent) sections of the file.

Quote:
Originally Posted by kolamorx View Post
What free tool for Windows do you recommend for volume measurement, so I can get data about the dialogue volume levels in some scenes per movie (which sometimes are very low), in order to set the maximum gain factor as less value as possible?
This tool, for example, can create a graph of the "loudness" (ITU BS.1770 and CBS) of a given file:
https://www.orban.com/meter/
__________________
There was of course no way of knowing whether you were being watched at any given moment.
How often, or on what system, the Thought Police plugged in on any individual wire was guesswork.



Last edited by LoRd_MuldeR; 7th May 2018 at 21:00.
LoRd_MuldeR is offline   Reply With Quote
Old 8th May 2018, 12:24   #162  |  Link
kolamorx
Registered User
 
Join Date: Jun 2017
Posts: 25
Quote:
Originally Posted by LoRd_MuldeR View Post
No matter what you do, the gain factor for each frame will never be allowed to exceed the frame's local max. gain factor – the maximum that this particular frame could be amplified without any clipping (distortion).
I know it, that's not what I meant.
Let me put it this way: if I set the maximum gain factor to too high value for a certain movie, sounds like breathing could have too high volume and that's a problem.
On the other hand, if I set the maximum gain factor to too low value for a certain movie, the dialogue volume could be too low comparing to the sound effects and music volume and that's a problem too.
Am I right?

Last edited by kolamorx; 8th May 2018 at 12:34.
kolamorx is offline   Reply With Quote
Old 8th May 2018, 18:53   #163  |  Link
LoRd_MuldeR
Software Developer
 
LoRd_MuldeR's Avatar
 
Join Date: Jun 2005
Location: Last House on Slunk Street
Posts: 13,032
Quote:
Originally Posted by kolamorx View Post
I know it, that's not what I meant.
Let me put it this way: if I set the maximum gain factor to too high value for a certain movie, sounds like breathing could have too high volume and that's a problem.
On the other hand, if I set the maximum gain factor to too low value for a certain movie, the dialogue volume could be too low comparing to the sound effects and music volume and that's a problem too.
Am I right?
Nope, not really.

The DynamicAudioNormalizer will amplify each part of the file, which is quieter than the target volume, so that it is brought up to the target volume. Also it will attenuate each part of the file, which is louder than the target volume (if any such part exists), so that it is brought down to the target volume. In the end, all parts of the file will have "equal" volume. Of course, that all combined with Gaussian smoothing in order to avoid fast fluctuation of the gain factors.

Now, what the "maximum gain factor" option really does: It imposes an additional limit on how much a "quiet" frame may be amplified in order to reach the desired target volume. The default "maximum gain factor" value, which is 10.0×, should be enough for most needs. But it can, of course, be increased, if it should really be needed. Anyhow, probably the vast majority of your frames (if not all of them!) will not need more amplification than what the default "maximum gain factor" value allows. And again: This is only an upper limit, not the actual amount of gain that will be applied! The actual gain factor will be determined dynamically for each frame, and it probably will be way lower than the limit in most cases!

The "maximum gain factor" limit exists primarily in order to avoid humongous amplification factors in extremely quiet (almost silent) sections of the file. If there was no such limit, it could easily result in excessive amplification factors in those extremely quiet sections – which you do not want. If we'd amplify extremely quiet (almost silent) sections by such an excessive amount, the result will sound very "noisy", because the quantization noise becomes audible...

(I suggest you let DynamicAudioNormalizer write a log file and inspect that log file in DynamicAudioNormalizer GUI, which will easily show whether you run into the limit or not)
__________________
There was of course no way of knowing whether you were being watched at any given moment.
How often, or on what system, the Thought Police plugged in on any individual wire was guesswork.



Last edited by LoRd_MuldeR; 8th May 2018 at 19:07.
LoRd_MuldeR is offline   Reply With Quote
Old 19th December 2018, 14:43   #164  |  Link
JuliusJac
Registered User
 
Join Date: Dec 2018
Posts: 2
ffmpeg dynaudnorm with almost no effect

Hei,

I know this Thread is rather old. But I have been searching the internet, and tried myself for 2 full days with almost no result. We have several hundred videofiles where the "volume" changes within the soundtrack. And I was trying all/a lot. But no loudnowm or dynaudnorm setting I use seems to giv an acseptable result.? can anyone Help me?

Here is one of the arguments I tried:

$ ffmpeg -i testin.mp3 -af "dynaudnorm=p=0.8:m=25:r=0.5"

Here one of our examples
testin.mp3

THANKS and Greetings from Norway
JuliusJac is offline   Reply With Quote
Old 19th December 2018, 15:50   #165  |  Link
LoRd_MuldeR
Software Developer
 
LoRd_MuldeR's Avatar
 
Join Date: Jun 2005
Location: Last House on Slunk Street
Posts: 13,032
Quote:
Originally Posted by JuliusJac View Post
Hei,

I know this Thread is rather old. But I have been searching the internet, and tried myself for 2 full days with almost no result. We have several hundred videofiles where the "volume" changes within the soundtrack. And I was trying all/a lot. But no loudnowm or dynaudnorm setting I use seems to giv an acseptable result.? can anyone Help me?

Here is one of the arguments I tried:

$ ffmpeg -i testin.mp3 -af "dynaudnorm=p=0.8:m=25:r=0.5"

Here one of our examples
testin.mp3

THANKS and Greetings from Norway
Hello.

Well, what are you trying to achieve? Or what is wrong with DynAudNorm default settings?

DynAudNorm was designed to retain the "dynamics" that exist in each local neighborhood, so that the characteristics of the original audio are preserved. Default settings have been chosen rather conservative.

Anyway, if you want to force DynAudNorm to act more "aggressively", first try decreasing the filter size ("--gauss-size" option). Default is 31, but you may want to try something like 11.

In addition to that, you may also try decreasing the frame length ("--frame-len" option). Default is 500 ms, but you could try 250 ms or even 125 ms. Note that filter size is expressed in frames, so these option mutually influence each other!

(I'm writing about the DynAudNorm CLI options here, because I'm more familiar with that one than with the FFmpeg port, but things should work for FFmpeg just the same way)


Your original:


DynAudNorm default settings:


More aggressive settings (filter size reduced to 11):


Something extreme (filter size reduced to 5 + frame length reduced to 125):
__________________
There was of course no way of knowing whether you were being watched at any given moment.
How often, or on what system, the Thought Police plugged in on any individual wire was guesswork.



Last edited by LoRd_MuldeR; 20th December 2018 at 13:18.
LoRd_MuldeR is offline   Reply With Quote
Old 20th December 2018, 09:53   #166  |  Link
JuliusJac
Registered User
 
Join Date: Dec 2018
Posts: 2
Hi LoRd_MuldeR

Thanks for your time!
I tuned the values in to the opposit direcction
for instance in this audiofine we have the periode 43-54sec and 54-69sec where the volume is is compleatly different. (manly due to the fact of a complete differnet sound)
Is there a way of eventing out this differance evan more?
JuliusJac is offline   Reply With Quote
Old 20th December 2018, 12:57   #167  |  Link
LoRd_MuldeR
Software Developer
 
LoRd_MuldeR's Avatar
 
Join Date: Jun 2005
Location: Last House on Slunk Street
Posts: 13,032
Quote:
Originally Posted by JuliusJac View Post
for instance in this audiofine we have the periode 43-54sec and 54-69sec where the volume is is compleatly different. (manly due to the fact of a complete differnet sound)
Is there a way of eventing out this differance evan more?
Well, in theory, you can go as low as "--gauss-size 3" and "--frame-len 10" – but be aware that, at some point, audio quality is going to suffer severely!

You will have to do some experiments in order to find the "sweet spot" for you needs

In addition to that, you may try the "--compress" option, which is going to prune the "loudest" samples prior to the actual normalization. Note that, here too, smaller parameter values result in stronger compression.

(Be aware: extreme compression values are going to distort the audio quality!)

Quote:
Originally Posted by JuliusJac View Post
I tuned the values in to the opposit direcction
See also:
http://muldersoft.com/docs/dyauno_re...#configuration
__________________
There was of course no way of knowing whether you were being watched at any given moment.
How often, or on what system, the Thought Police plugged in on any individual wire was guesswork.



Last edited by LoRd_MuldeR; 20th December 2018 at 15:14.
LoRd_MuldeR is offline   Reply With Quote
Old 14th July 2019, 20:08   #168  |  Link
Emulgator
Big Bit Savings Now !
 
Emulgator's Avatar
 
Join Date: Feb 2007
Location: close to the wall
Posts: 717
I just want to say thank you and give it a test:
But where am I stuck ?
On startup the DAN GUI.exe wants to see the path to a logfile that doesn't exist yet.
Any option I press then leads to closing of the program.
I faked a logfile to get into it and probably access settings, but: it opened the logviewer, if I close it: no GUI open anymore.
__________________
"To bypass shortcuts and find suffering...is called QUALity" (Die toten Augen von Friedrichshain)
"Data reduction ? Yep, Sir. We're working on that issue. Synce invntoin uf lingöage..."
Emulgator is offline   Reply With Quote
Old 14th July 2019, 21:12   #169  |  Link
LoRd_MuldeR
Software Developer
 
LoRd_MuldeR's Avatar
 
Join Date: Jun 2005
Location: Last House on Slunk Street
Posts: 13,032
Quote:
Originally Posted by Emulgator View Post
I just want to say thank you and give it a test:
But where am I stuck ?
On startup the DAN GUI.exe wants to see the path to a logfile that doesn't exist yet.
Any option I press then leads to closing of the program.
I faked a logfile to get into it and probably access settings, but: it opened the logviewer, if I close it: no GUI open anymore.
Sorry, the GUI program is just to show (visualize) existing log files:
http://muldersoft.com/docs/img/dyauno/Chart.png

You have to use the CLI program for processing audio data; it can optionally produce a log file:
http://muldersoft.com/docs/dyauno_re...and-line-usage

If you want a GUI, you are probably better off by using the VST plug-in:
http://muldersoft.com/docs/dyauno_re...-plug-in-usage
__________________
There was of course no way of knowing whether you were being watched at any given moment.
How often, or on what system, the Thought Police plugged in on any individual wire was guesswork.



Last edited by LoRd_MuldeR; 14th July 2019 at 21:18.
LoRd_MuldeR is offline   Reply With Quote
Old 15th July 2019, 23:12   #170  |  Link
Emulgator
Big Bit Savings Now !
 
Emulgator's Avatar
 
Join Date: Feb 2007
Location: close to the wall
Posts: 717
Ah thanks, that clears it up.
Munching...
Audacity 2.3.2. sees, but refuses to load DAN VST.dll
SoundForge11 does not see DAN VST.dll

At only 42KB the .dll is quite small. Missing dependencies ?

Solved: I should have used the shared .dll of course. Now both DAWs can see and use DAN.

Awsome work, Lord Mulder !
Just applied it to the sound track of a just restored 1961 movie using the defaults.
Gets the (normally tedious manual) work beautifully done, keeping the quiet/loud ratio convincing, no pulling gain up in quiet parts.
__________________
"To bypass shortcuts and find suffering...is called QUALity" (Die toten Augen von Friedrichshain)
"Data reduction ? Yep, Sir. We're working on that issue. Synce invntoin uf lingöage..."

Last edited by Emulgator; 16th July 2019 at 08:20.
Emulgator is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 04:28.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.