Do different audio settings result in different loudness?

[Filmriss]

Hi,

I noticed, that x264 files from the internet tend to sound quiet and as if someone is speaking into his hands on my TV.

I have converted a file myself and set the audio to "auto-passthrough", but afterwords the audio did sound not as loud as before on my Computer. Also the format changes from mpeg to AC3 and from 265kbit/s 8bit to 640kbit/s 16bit. The second is the quieter one.

[rollin_eng]

Could you please post your logs, instructions can be found here:

viewtopic.php?f=6&t=31236

[Woodstock]

Then your audio was not one of the audio codecs that you enabled for "auto passthrough", so it was converted. So yes, the audio levels can change.

[mduell]

Then your audio was not one of the audio codecs that you enabled for "auto passthrough", so it was converted. So yes, the audio levels can change.

If only there was some way to see what actually happened without relying on user's unreliable descriptions.

If there was, they could make it mandatory in all posts asking for help.

Wouldn't that be useful?

[Filmriss]

SORRY, I thought I posted them here as well. (I have another thread, different topic focus, where I posted it).

Pastebin does not work, which is why I always first see, if the problem can be solved without uploading the log.

The log is here: https://www.dropbox.com/s/y5ub0g26x2p6s ... g.rtf?dl=0

[Filmriss]

In case I have not chased away everyone who could have helped me, did anyone see something in the LOG that is the reason why the audio is always quieter after encoding?

[Woodstock]

It would have been a bit simpler if you had posted the text, rather than an RTF file... Every time I tried to move around in the file, document navigation bars popped up, and the page breaks confused a few things.

Your video isn't even using MP3 audio, so the first two tracks have to be converted. You may have better results converting your selected track to MP3. AC3 audio is often "compressed", so it seems louder. I do not know if handbrake does that type of audio compression when converting to AC3.

You CAN tell handbrake to adjust the volume, if you click on the "show" button on the audio track in the graphic user interface. That opens up additional options, including a gain setting.

[Filmriss]

The scrollbar popping up might come, if you watch it on Dropbox instead of first downloading it. Otherwise sorry, I didn't know this could be an issue, since in my rtf viewer on my computer (not dropbox) scrolling works and there are no jumps. I'll post it into the thread next time.

About the file:

I am a bit confused. Handbrake offers me 3 audio tracks, see picture (MediaInfo shows two mpeg tracks, Handbrake screenshot shows 3 audio tracks). Two of these seem to be mpeg1 and one AC3.

I don't remember which of the countless trys is the one that you said I had converted to AC3. I thought I went with AAC preferably and also tried others. ALL AC3, AAC, MP3 and passthrough were QUIETER than the original mpeg audio track from the .ts file.

You say AC3 can seem louder. Are you saying this in regard to the source? Then VLC must play back the AC3 track in the source file (the .ts file). Because the tracks of the converted movie are the ones that are quieter. My problem is, that the result is quieter than the source

I would upload the pictures in the thread, so it was easier for you, but there is no attachment button here so I have to use Dropbox again, sorry. https://www.dropbox.com/sh/c5z9to12pq0z ... ODLWa?dl=0

[Filmriss]

You CAN tell handbrake to adjust the volume, if you click on the "show" button on the audio track in the graphic user interface. That opens up additional options, including a gain setting.

I don't find the "show" button, but I knew that there is a "gain" knob under the audio column. I assume "gain" is loudness, but I always kept it to zero, thinking it would result in a 1:1 copy of the loudness.

[Filmriss]

MAYBE THE SOLUTION:

I found out, that

QUICKTIME:
- plays ts-source file loud (Quicktime info windows shows 3 audio tracks, 2x mp3, 1x AC3)
- doesn't play mp3
- plays passthrugh, AAC and AC3 quieter

VLC plays:
- all files at the same loudness level
- .ts is quieter than in Quicktime, but the same level as the others
- also plays mp3 (I mean the mp3 that is in the resulting file)


My assumption, Quicktime maybe plays all 3 tracks at the same time?

[Woodstock]

I keep forgetting that the Windows and Mac GUIs are so different... The Windows GUI hides a lot of lesser-used options, like the gain.

It is doubtful that Quicktime would play more than one track at a time, but its decoders can have different bias towards different encodings. AC3 uses more audio compression that most other encoders, which makes it seem louder. Kind of like the audio level difference between programs and commercials on TV. Used to be, the volume level stayed nominally the same, but the audio was compressed to make it stand out on the advertising.

[Filmriss]

Hm, yes, but when you say AC3 seems louder, why is the AC3 in the resulting file (after encoding) quieter? Shouldn't it be the other way round?

Source: ........................... Resulting file
3tracks (2x MP3, 1x AC3)....... 1 track AC3
loud ............................... quiet

Hm, ok, it's probably not worth wondering, I was just curious. I can of course just use the gain button.

Thank you Woodstock!

I think the other members avoid me, since I didn't upload a log from the beginning in two other threads. I don't expect the other two threads to be answered.

May I ask you, is it allowed on this forum to upload sample files and ask, which one is better (I can't judge with my eyes), or is it likely that no one will answer since picture quality is subjective to the eye of the beholder? Or maybe such questions are condoned in general?

[Woodstock]

I can't speak for other members. Posting logs is VERY important whenever asking, "What should I change?" or "Why doesn't this work?" questions. Sometimes it doesn't help, but that's rare, hence the emphasis on it in the support areas, to the point that clicking on "Post a message" comes up with a template asking for the logs. Your question really falls into support, rather than general chit-chat. So logs showing what you're working with, and what track numbers were assigned during the conversion, are very helpful.

Samples with the logs are usually better than just the logs, but samples without logs aren't very helpful.

A lot of times, people say, "But the log isn't relevant!", then finally post their log, and the answer is found immediately, and it wasn't something the poster thought had any bearing on the matter.

Information from your playback software as to which track it decided to play is important, too; quite often, it is NOT the track you THINK is being played! Players have preferences as to which tracks they play by default. Sometimes, it is the one you flag as default. Sometimes, it isn't.

My experience is that, if the source has an AC3 track and handbrake is told to convert it to an AAC track AND pass the AC3 track through "untouched" as a second track, when comparing the AAC and AC3 track on the same playback hardware, the AC3 track seems considerably louder. I have not tried it with MP3 in the conversion mix, so I can't say much about the relative volume of those.

[Filmriss]

I am uploading the logs and the audio test files I prepared two days ago, to dropbox and will post a link. The folder will include every audio "from a to b" combination that is possible. I always selected the mpeg track and encoded that to the selectable output possibilities.

*ts is the source file. Again the source (the ts file) is louder than the readily encoded files!

I will post the link, once it is ready. (you will also see that I was not lying/claiming about having tried something, like mduell implied).

Playback software: I don't know how to see, what track Quicktime uses during playback. Hm, when I look into VLC it lists all three tracks in the source file when playing, too. But I guess that is not telling, what it actually has selected for playback.

PS: I see I did already upload one log above, which you said is saying that there isn't even basic mp3 in the source.

Edit: Dropbox says 1h. I will have a look, if I am still awake, when it has finished uploading. It is 2h in the night here.

[Filmriss]

Here is the folder including all selectable audio output formats. https://www.dropbox.com/s/03atoqu594c9s ... 0.zip?dl=0
and the log.

[Woodstock]

I am certainly not an expert at this. However, playing all of samples you provided in VLC, "back to back", the audio level seems identical between the original and the AAC.

[Filmriss]

Yeah, I found out levels are the same in VLC, but in Quicktime they differ. Strange. There is also no way to adjust audio level in Quicktime. Well, I have to take it as it is, I think.

[arnold2]

One of the reason for the apparent change in 'levels' is the amount of compression applied to the audio file.

Often big action films have the dialogue set to a very low volume, but the big explosions etc. sounds set to be very loud.

Compression squashes the audio levels, so quiet stuff becomes nearer the same volume as the loud ones, making dialogue clearer.

AAC, especially when set to a low bitrate of 128Kb or less, squashes up the audio as a function of trying to save file size space - hence why it may 'sound louder'.

However, re-encoding from 128K up to say 700k won't 'uncompress' the file!

[Deleted User 11865]

I believe you're talking about dynamic range compression. It's not linked to bitrate at all (generally speaking, I can't speak for hacks implemented in this or that encoder), and some audio compression formats define a way to store DRC hints when encoding (so that DRC is NOT applied at all by the encoder, rather, the decoded reads the hints and applies them - most decoders let you disable it).

As far as HandBrake goes, our audio decoders ignore all DRC hints, unless tell it to (by setting DRC to a non-zero value). We currently only support applying DRC from AC-3 and E-AC-3 sources (so that excludes e.g. "MPEG" audio).

In addition, our audio encoders currently do not support writing DRC hints to the bitstream. So unless you use audio passthru, re-encoding audio with HandBrake effectively erases any DRC hints present in the input audio track.

So yeah, some "loudness" differences could be explained by the fact that you were playing the source on a device that supports hint-based DRC and has it enabled, whereas HandBrake's output has no such hints (effectively disabling DRC).That's only one scenario though.

[Filmriss]

Thanks arnold2,

the same as Rodeo, I am rather reminded of DRC with what you are talking. Not the issue I am having.

And, yes of course, using a higher bitrate for a file already having a small bitrate within the same codec doesn't make it uncompressed. I mean you can't add, what is not there any longer. (However I learned that there is a different case, when you have e.g. 256kbps MP3 and then convert it to 256kbpy AAC, you might want to choose a higher bitrate for AAC, to preserve more of the already lossy MP3. -> instead of having a smallfraction of something that has already fractioned, you keep a bigger fraction of it.)

@Rodeo
Thanks, but I am aware of DRC (I know, you couldn't know). I got aware of it a long time ago, when I was anoyed that my Star Wars DVD's sound was fluctuating all the time, when there was music or speech etc.

The issue above doesn't seem to have to do something with this.
1. I allready had looked into Quicktime, if it was probably had DRC enabled and I didn't see, it was enabled
2. I had the DRC value in Handbrake set to 0. (You asked for the log and that is probably in th elog I uploaded with the dropbox-link. PS: I in the meantime deleted all the files, because I didn't want to keep it there any longer and let the movie files waste my small 2GB account)
3. as I listed, VLC plays all at the same level, QUicktime only plays the original mpeg (1 or 2) audio loud, but doesn't play mp3 from Handbrake at all, and plays AC3 and AAC quieter than the original mpeg (1 or 2).
4. I should have been clearer and say, that the audio levels don't change within one track, i.e. there is no fluctuation in the level, but the tracks compared to each other don't have the same level.

PS: when I used words like loudness or noise,above, I used them in a colloquial meaning (or probably in wrong english), but not in the meaning of a terminus technicus, like in "loudness war" and noise describing hizzing etc. Sorry, for that, I might have been the reason, why you confused it with DRC. Calling the level "loudness"