View Full Version : 6 Channels to Stereo
doomleox999
25th August 2016, 23:26
Hi everyone, english is not my first language, so please excuse any mistakes.
I use MeGUI and when I encode 6 channels audio to stereo, the volume of the final result is lower.
This are the properties (?) of the 6 channels audio:
http://i.imgur.com/fjpEQw6.png
And this are the settings of the desired encoded audio:
http://i.imgur.com/igcUpL6.png
Thanks in advance.
tebasuna51
26th August 2016, 10:06
- Please put the full log of your conversion.
- I can't understand the audio properties of your input file: Format profile, Channels_original, Channel position, ...
There are something wrong in that.
- Never use Directshow decoder like first option, use always first LWLibavAudioSource
- Using Normalize Peaks to 100% you have the max volume without distort.
You have:
1) Overall low volume?
or
2) Low dialog volume?
If 1) maybe there are something wrong in your audio input.
If 2) maybe you need a special downmix of your source.
doomleox999
26th August 2016, 11:45
- Please put the full log of your conversion.
- I can't understand the audio properties of your input file: Format profile, Channels_original, Channel position, ...
There are something wrong in that.
- Never use Directshow decoder like first option, use always first LWLibavAudioSource
- Using Normalize Peaks to 100% you have the max volume without distort.
You have:
1) Overall low volume?
or
2) Low dialog volume?
If 1) maybe there are something wrong in your audio input.
If 2) maybe you need a special downmix of your source.
I will put the log in my next reply.
I get those properties with MediaInfo. Do you think there is something wrong with the program?
Ok, I'll start using LWLibavAudioSource from now on.
I have normalize to %100 already checked.
I get overall low volume.
tebasuna51
26th August 2016, 13:42
I get those properties with MediaInfo. Do you think there is something wrong with the program?
We can see only Audio info, for that we need the full log, we don't know if this AAC is inside a container or a standalone audio file.
A aac stream must be HE-AACv2 or HE-AAC or LC but not the 3 profiles at time, a HE-AACv2 can't have 6 channels, what means 20 channels (Front: 12, Side: 7, Back: C)...
There are something wrong in MediaInfo or in your audio stream.
Ok, I'll start using LWLibavAudioSource from now on.
Yep, put the full log using LWLibavAudioSource, DirectShow can modify the number of channels regarding your PC audio configuration.
doomleox999
30th August 2016, 04:49
Ok, so first of all, these are my desired encoding settings (audio):
http://i.imgur.com/H8jBQRC.png
doomleox999
30th August 2016, 04:54
These are the .mkv file properties (video+audio):
http://textuploader.com/58u0k
And this is the log of the encode:
http://textuploader.com/58u00
tebasuna51
30th August 2016, 11:11
1) Please verify than you use:
Options -> Settings -> Extra Configuration -> Auto Update -> Use Development Update Server
2) And update your soft
Options -> Update
Because:
--[Information] [29/08/2016 07:51:41 p.m.] Automatic update is disabled
--[Information] [29/08/2016 07:51:41 p.m.] Using cached update config and server: http://megui.org/auto/stable/
--[Information] [29/08/2016 07:51:42 p.m.] There are 8 packages which can be updated: MeGUI, MeGUI Libraries, MediaInfo, x264, x265, MP4Box, FFMS, L-SMASH Works
Now your audio stream can be decoded only with DirectShowSource:
--[Information] [29/08/2016 07:55:52 p.m.] Trying to open the file with LWLibavAudioSource()
--[Information] [29/08/2016 07:55:53 p.m.] Failed opening the file with LWLibavAudioSource(). LWLibavAudioSource: failed to get the audio track.
--[Information] [29/08/2016 07:55:55 p.m.] Trying to open the file with FFAudioSource()
--[Information] [29/08/2016 07:57:59 p.m.] Failed opening the file with FFAudioSource(). FFAudioSource: Failed to open 'C:\MOVIE.mkv.ffindex'
--[Information] [29/08/2016 07:57:59 p.m.] Trying to open the file with DirectShowSource()
--[Information] [29/08/2016 07:58:04 p.m.] Successfully opened the file with DirectShowSource()
--[Information] [29/08/2016 07:58:04 p.m.] Avisynth script
---[NoImage] LoadPlugin("C:\Program Files (x86)\MeGUI_2507_x86\tools\avs\plugins\directshowsource.dll")
---[NoImage] DirectShowSource("C:\MOVIE.mkv", video=false)
And we don't know if the audio is decoded properly or not.
3) BTW, now the MediaInfo of your audio is coherent:
---[Information] Audio
----[Information] FormatProfile: HE-AAC / LC
----[Information] FormatSettingsSBR: Yes (Implicit)
----[Information] SamplingRateString: 48.0 kHz / 24.0 kHz
----[Information] Channels: 6
----[Information] ChannelPositionsString2: 3/2/0.1
If your audio is still with low volume try after update MeGUI, until the file can be oppened by LWLibavAudioSource().
4) There are also other recomendations to avoid problems with folders protected by Windows:
a) Don't install portable soft, like MeGUI, in: C:\Program Files (x86)\MeGUI_2507_x86\...
Use a folder created by the user, for instance: C:\Portable\MeGUI...
b) Don't put data in root folder: C:\MOVIE.mkv
Use a folder created by the user, for instance: C:\Data\MOVIE.mkv
hello_hello
30th August 2016, 16:20
Some thoughts....
Multichannel audio might sound a bit quieter than the original after it's been downmixed to stereo, but that's probably relative. If you're listening to the original audio downmixed to stereo on playback maybe it's being downmixed differently.
MeGUI uses a standard formulae for downmixing to stereo (the volume is reduced enough to prevent clipping when the channels are combined) then if the "normalise" option is selected it's adjusted so the peaks are at maximum (assuming you have the option set to 100%).
What happens if the same multichannel audio is simply downmixed to stereo on playback though.... and there's no volume reduction to prevent clipping? It could sound louder than the downmixed version created by MeGUI, but the peaks could also be clipped. Or sometimes the downmixed version created by MeGUI might sound louder, or sometimes about the same, it'd depend on the dynamic range (how loud the peaks are compared to the average volume) and how the original is being downmixed on playback.
Also... I've found after downmixing, the volume often seems to be similar to the original on playback. Not always, but it's rarely low enough to concern me, however that's assuming the multichannel audio you're downmixing and encoding is the original audio. If it's the original AC3 or DTS multichannel, the peaks might be a fair way below maximum, so after downmixing it and normalising, the stereo version could have a similar volume, but if your source is multichannel audio that someone else normalised when they encoded it (ie if the original multichannel AC3 was normalised when it was encoded as multichannel AAC) then the AAC version is probably louder than the "original" AC3, and it's far more likely your downmixed version will sound quieter by comparison.
doomleox999
30th August 2016, 23:59
Thanks all for your answers and information, the more you know, the better.
I don't really think that the instalation folder of MeGUI is important because I never had a problem, and the volume loss when downmixing only happens with certain files, right now it only happens with this one. I have another .mkv file with a 6 channels audio that maintains the same volume when downmixing.
I moved the file to the desktop and encoded again, but I got the same result.
Here is the log: http://textuploader.com/58uzi
doomleox999
31st August 2016, 04:14
Is it against the rules if I upload a 1 minute sample of the .mkv file (video+audio), so you can test it and/or encode it with your computer, to see if you have the same problem?
tebasuna51
31st August 2016, 12:57
If is a downloaded mkv, that you do not own the sources, we can't help you.
doomleox999
31st August 2016, 16:48
Ok then, I'm not sure about the "owning the sources" thing but it doesn't matter.
It's not a big deal, I'll bare the volume loss and that's it. Thanks for all the help.
hello_hello
1st September 2016, 00:27
Try encoding it as a wave file.... well if you use MeGUI it'll have to be flac for lossless..... convert the flac file to wave and open it with an audio editor (Audacity or something similar). Most audio editors have a normalise function. If you use it and the volume doesn't increase (it won't) then the peaks were already at maximum.
Or scan it with ReplayGain. Foobar2000 can ReplayGain scan anything it can decode, which is most audio formats (third party plug-ins can be downloaded to decode AC3 and DTS. The scan result will include the peak levels. "1" is maximum. Sometimes it'll be a little more.... 1.04562 or something similar.... sometimes a little less, but it should be very close to "1" if it's been peak normalised.
I use foobar2000 for most of my audio encoding. For AAC in an M4A or MP4, after a scan foobar2000 can use the result to adjust the volume losslessly (same for MP3). Because lossy encoders can store audio above "0" or maximum for a wave file, you could increase the volume a little more that way and make the peaks +3dB to squeeze out a tad more volume. I probably wouldn't increase it much further than that, but it can be done. Losslessly for AAC and MP3. Or you could re-encode the audio while compressing it to reduce the peaks levels. foobar2000 comes with a couple of limiter DSPs you can use when converting and it also has a downmixing DSP, although I'd recommend using the Matrix Mixer DSP instead as it automatically normalises and you can configure it any way you like.
After a ReplayGain scan a peak value of roughly 1.4 is a peak of +3dB.
raffriff42
1st September 2016, 00:52
@hello_hello, it seems Normalise is already selected (post #5). I wonder if the source has out of phase channels, causing phase cancellation & low volume when mixed together. Or a very loud LFE channel, pushing down the level of the overall mix when normalised.
tebasuna51
1st September 2016, 11:31
I wonder if the source has out of phase channels, causing phase cancellation & low volume when mixed together. Or a very loud LFE channel, pushing down the level of the overall mix when normalised.
I agree with your 2 reasons, even a little error in decoder can produce peaks than prevent than Normalize work.
Sometimes a very simple downmix like:
a=LWLibavAudioSource("D:\tmp\sample5.1.aac")
f = GetChannel(a, 1, 2)
c = GetChannel(a, 3, 3)
MixAudio(f, c, 0.3)
can work better (we can test other values than 0.3 for mix)
hello_hello
1st September 2016, 13:16
@hello_hello, it seems Normalise is already selected (post #5). I wonder if the source has out of phase channels, causing phase cancellation & low volume when mixed together. Or a very loud LFE channel, pushing down the level of the overall mix when normalised.
Good thoughts.
I never include the LFE channel when downmixing myself and I reduce the volume of the surround channels 3dB relative to the front channels. Only because I prefer it that way. Any headroom gain is just a bonus.
I stopped including the LFE channel because I downmix to stereo and compress a little on playback (night mode) and I found the only time I noticed the compression in action was occasional places with high levels of low frequency content, which would cause the volume of speech to drop. It almost never happens when the LFE isn't included.
MeGUI includes the LFE (for stereo downmixing, but not Prologic) and the surround channels aren't reduced. I don't think there's any way to change that, although you could manually create a script to open the audio, add your preferred stereo downmixing, and load the script into the audio section for encoding. (Edit: I just noticed tebasuna51 has already posted one).
Or there's an AVS Input DSP for foobar200 that'll allow it to open and encode scripts too.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.