Mini Guide: Compress dynamic range stereo audio for home use using Audacity

[nbarzgar]

Hello,
I post this here, because I found my own little method based on hints here at the forum. For anyone interested to not have their ear drums split and also be friendly to their neighbours' nerves and sleeping hours while watching movies, I share this here as a short step-by-step guide.

Short background:
As others have posted on this fourm already, movies tend to be created for the home theatre enthusiast who watches in broad daylight, perhaps with friends, family, and the said neighbours. But sometimes you watch at night and although many modern devices (Blu-ray player, TV, you name it) have dynamic compression settings included, they mostly are not sufficient to compress what is being delivered on DVDs and Blu-Ray disks as 6.1 or even more channel audio and meant to give the true cinema feeling - which means sounds and music are deafeningly loud, speech on the other hand so low you have to continually adjust the volume - or wear earphones... Both not overly satisfying solutions.

So here goes:

• Convert any audio to stereo first (look around the forum, I always use BeSweet with BeLight GUI).
• Save the file as .wav (makes opening in Audacity faster, if necessary).
• Open the stereo .wav file in Audacity.
• From the 'Effects' menu, choose 'Compressor'.
• Use default settings, except for the first option in the list, as shown in the screenshot (says something like 'limit' or 'limiter'; note: screenshot taken from the German version of Audacity).
• Push the slider of that option to the extreme left, as shown, so the curve in the little diagram is almost level with the upper box margin.
• Let Audacity do its work and take care that as much as possible of the 'dynamics' (i.e. different wave heights, simply put) are levelled out.
• Save the compressed audio to a new .wav file, then convert to .mp3 or .ac3.
  (Found that more reliable, than saving directly to the new audio format.)

Audacity Compressor setting:


Sample view of audio before and after applying compressor:


As visible in the screenshot this method does not level out all dynamics completely, which one wouldn't want. But sliding the switch of that option in Audacity to somewhere on the extreme left takes care of a lot.
EDIT: Adding this note: Keep the original audio, in case you want to use the multi-channel layout someday or somewhere. EDIT
Of course you may need to try this for a few times until you reach the level of compression best suited to your ears and environment.
Hope this helps.

[sneaker]

Convert any audio to stereo first (look around the forum, I always use BeSweet with BeLight GUI).

Is this really a good idea? Wouldn't it make more sense to down-mix with speech boosting (center channel) somehow (if necessary)? Once they are mixed you can't fix effects being louder than speech.

[nbarzgar]

Convert any audio to stereo first (look around the forum, I always use BeSweet with BeLight GUI).

Is this really a good idea? Wouldn't it make more sense to down-mix with speech boosting (center channel) somehow (if necessary)? Once they are mixed you can't fix effects being louder than speech.

Hello sneaker, well, I

• am mainly about the simplest solution possible and
• would presume someone reading this is reasonably familiar with techniques - so I assume they keep the original audio - but added a note to my post, to make sure
• and also think it's about making it well-audible on a TV (stereo output expected)
• while if you use a full-blown home theatre device setup, you would keep channels
• and in that case, more effort and a less simple procedure would be needed I haven't tried yet.
• That's why it says in the title "stereo... for home use" to make clear it's meant to be that simple.
• The channels themselves, if converted to stereo, usually look like the 'before' stream in the sample screenshot I included, therefore 'levelling out' those high dynamic ranges makes sense. In my case.

But thanks for the contribution!
If you know how to do what you propose, feel free to write another thread in this forum about it.
(Would be lost here for people to find, if you added it here, I think.)

[video.baba]

[*]am mainly about the simplest solution possible ...

Really? Let's re-cap....

• Convert any audio to stereo first (look around the forum, I always use BeSweet with BeLight GUI).
  Save the file as .wav (makes opening in Audacity faster, if necessary).
  Open the stereo .wav file in Audacity.
  From the 'Effects' menu, choose 'Compressor'.
  Use default settings, except for the first option in the list, as shown in the screenshot (says something like 'limit' or 'limiter'; note: screenshot taken from the German version of Audacity).
  Push the slider of that option to the extreme left, as shown, so the curve in the little diagram is almost level with the upper box margin.
  Let Audacity do its work and take care that as much as possible of the 'dynamics' (i.e. different wave heights, simply put) are levelled out.
  Save the compressed audio to a new .wav file, then convert to .mp3 or .ac3. (Found that more reliable, than saving directly to the new audio format.)

Or you could just load the file into BOX4 and press 'START'.

[nbarzgar]

video.baba
Or you could just load the file into BOX4 and press 'START'.

Of course, you can.
But I've tried a lot of 'out-of-the-box' solutions and since I am not satisfied with the results, I rather demux audio, convert it, then remux. It's simple really, if you like doing this.
If not: Do it your way.

[hello_hello]

nbarzgar,
I've never used BOX4 but I see it uses the Dynamic Audio Normaliser which is quite good so I'd recommend trying it. I don't know how much control BOX4 gives you over the Dynamic Audio Normaliser settings though.
There's nothing wrong with downmixing, compressing and converting with different programs, but doing it all in a single step if possible is less tedious.

Dynamic Audio Normaliser is a command line utility and it's not particularly GUI friendly, however ffmpeg has it built in these days.
My preference is to use foobar2000's converter, and I find the Dynamic Audio Normaliser works well although the defaults are a little slow to respond to sudden changes in volume. A DSP can be added to the foobar2000 conversion chain for downmixing if need be, and the whole configuration can be saved as a conversion preset.

A foobar2000/ffmpeg command line for compressing and converting to 128k MP3 with the Dynamic Audio Normaliser's "window size" reduced so it responds faster, or for a super-fast response you could use something like -af dynaudnorm=f=75:g=11.

-i - -ignore_length true -af dynaudnorm=f=150 -c:a libmp3lame -b:a 128k %d

You could include muxing by doing the whole lot with ffmpeg (I'm not familiar with ffmpeg's downmixing myself although I know it can downmix) and there's several GUIs you can use. The following ffmpeg preset for AnotherGUI should open most file types, copy the video, and compress and convert the audio to mp3 while remuxing as MKV.

-i "<FullSourceFileName>" -y -threads 1 -c:v copy -c:s copy -af dynaudnorm=f=75:g=11 -c:a libmp3lame -b:a 128k -compression_level 2 "<OutputPath><OutputFileName>.mkv"

[nbarzgar]

Hello @hello_hello
appreciate that too.
But here comes the limitation, I am facing:
I know a lot about the principles and using several applications - for years now, even in a chain. But I am not an audio technician or software developer. Therefore I would need quite some time to be able to creatively use and adapt your solution.
I've used foobar2000 a lot for all kinds for conversions and do still. But depending on the source, I like the options and straightforward approach of BeSweet and BeLight as GUI for audio conversion.
Where BeSweet GUI can appear cluttered and even confusing, BeLight is easy to use (if you know the basics), and it's also geared towards people using free software to do 'movie' editing, generally speaking. (Multi-channel audio, DNR, presets for dynamic range compression, convert to stereo, etc.)

And I like the visual control I have using Audacity, since I see what's happening to the waveform. Additionally I can easily use different settings to achieve the best results. But don't have to study the details of attack, release, etc.

[yilager]

It worked but I've got some sort of echo after doing this things. My settings are:
54
40
9:1
0.20
1.0

Is there any workaround?

[nbarzgar]

It worked but I've got some sort of echo after doing this things. My settings are:
54
40
9:1
0.20
1.0

Is there any workaround?

Hello @yilager, a lot depends on the source.
And the device you are playing back on...
Not sure also, what the 'echo' you mention really sounds like so for me it's hard to tell.

Alas, am a free-time user of these things; know quite a lot about video codecs, formats, resolutions, etc. Have done a lot of audio filtering and editing usually with free tools.
But I also always aim at getting things done so they sound well on my devices/setup.
I suggest, you change the values a little, play around and listen to the results until you find the settings that are fine for you.

Otherwise I'd strongly and kindly suggest you open a new thread in the forum. My post here is rather 'old', though still valid (for me). And your question might need more attention.
Best regards