Is Handbrake converting my audio or not?

[Foebane72]

I am using Handbrake to convert my videos to a smaller video size, but using "Auto Passthru" to hopefully leave the audio alone as the quality would drop too much if I also converted the audio. The recordings are AVC and AAC in origin, but for some reason MediaInfoXP is reporting the audio as different before and after. BTW, all the source videos are recorded by the same software (Nvidia Shadowplay).

Here is an example with one converted file:

Audio
ID                                       : 2
Format                                   : AAC
Format/Info                              : Advanced Audio Codec
Format profile                           : LC
Codec ID                                 : 40
Duration                                 : 13 min 43 s
Source duration                          : 13 min 43 s
Bit rate mode                            : Constant
Bit rate                                 : 192 kb/s
Channel(s)                               : 2 channels
Channel positions                        : Front: L R
Sampling rate                            : 48.0 kHz
Frame rate                               : 46.875 FPS (1024 spf)
Compression mode                         : Lossy
Stream size                              : 19.4 MiB (6%)
Source stream size                       : 19.4 MiB (6%)
mdhd_Duration                            : 823677

And what happens to the audio after, even with "Auto Passthru":

Audio
ID                                       : 2
Format                                   : AAC
Format/Info                              : Advanced Audio Codec
Format profile                           : LC
Codec ID                                 : 40
Duration                                 : 13 min 43 s
Bit rate mode                            : Variable
Bit rate                                 : 197 kb/s
Channel(s)                               : 2 channels
Channel positions                        : Front: L R
Sampling rate                            : 48.0 kHz
Frame rate                               : 46.875 FPS (1024 spf)
Compression mode                         : Lossy
Stream size                              : 19.4 MiB (15%)
Title                                    : Stereo / Stereo
Default                                  : Yes
Alternate group                          : 1
Encoded date                             : UTC 2016-10-12 08:28:23
Tagged date                              : UTC 2016-10-12 08:28:23

There is an obvious change, even though the audio sounds identical to me - so why is Handbrake converting my AAC audio when I don't want it to? Was I meant to set it to "AAC Passthru" instead? I'd hate to have to re-record more videos just to correct that!

[hello_hello]

I'd be willing to bet it's the same audio, maybe in a different container, and depending on the format MediaInfo is given different info and/or has to do some guessing. Each of these are the same audio. How it's reported by MediaInfo:

MKV without track statistic tags:
ID : 2
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : LC
Codec ID : A_AAC
Duration : 44 min
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 48.0 kHz
Frame rate : 46.875 FPS (1024 spf)
Compression mode : Lossy
Language : English
Default : Yes
Forced : No

MKV remuxed by MKVToolNix with track statistic tags:
ID : 2
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : LC
Codec ID : A_AAC
Duration : 44 min
Bit rate : 185 kb/s
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 48.0 kHz
Frame rate : 46.875 FPS (1024 spf)
Compression mode : Lossy
Stream size : 58.4 MiB (18%)
Language : English
Default : Yes
Forced : No

Remuxed as MP4 by ffmpeg:
ID : 2
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : LC
Codec ID : 40
Duration : 44 min
Bit rate mode : Constant
Bit rate : 185 kb/s
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 48.0 kHz
Frame rate : 46.875 FPS (1024 spf)
Compression mode : Lossy
Stream size : 58.4 MiB (18%)
Language : English
Default : Yes
Alternate group : 1

In your case the audio is reported as variable bitrate and constant bitrate, which still doesn't mean it's different, but it might warrant further investigation. Can you upload a small sample of the source and the Handbrake output?

[Foebane72]

Can you upload a small sample of the source and the Handbrake output?

Try these:
Before
After

I'm hoping the audio data in both will be absolutely identical in content.

[hello_hello]

I'm pretty confident they're the same. I remuxed both with MKVToolNix as MKA containing just the audio while splitting each at 19 seconds to make them the same length. I then extracted the raw AAC from each MKA with MKVCleaver. Explorer shows both as being the same file size (down to the byte) and foobar2000 shows the same number of samples. I converted both to 24bit flac with foobar2000 and the resulting MD5 for each was the same. If there were any differences they wouldn't be.

MP3DirectCut shows the AAC is encoded as average bitrate (199kbps). That's probably right.... the average bitrate part. The bitrate does vary a little.
Or AAC can use a "bit reservoir" even when it's encoded as constant bit rate which means unused bits in one frame can be used for the next one (lossy audio is stored in "frames")... or something like that... but it might cause audio encoded as constant bitrate to appear variable. I'm not 100% sure how all that works. I'm sure it's the same audio though.

[Foebane72]

Thank you very much for your efforts, Hello_Hello - it's much appreciated!

[jagabo]

I opened the two tracks in Audacity, split them into separate left and right tracks, delete the right tracks, inverted one of the left tracks and made it a right track, and joined them as a stereo track. Finally I rendered them to a mono track. Mathematically: (L1 + (-L2)) / 2. The result was zero, absolute silence. So the two left tracks are identical. I would assume the right tracks are the same too.

[Foebane72]

Thanks, Jagabo!

[Krispy Kritter]

I wouldn't have expected them to be different either as the stream size is the same for both. If it was converting, or changing from constant to variable, the stream size would be affected.

[hello_hello]

The only problem there is if you re-encode 192kbps constant bitrate audio as 192kbps constant bitrate audio they'll be the same bitrate and the same file size. The same could also apply when encoding constant bitrate as average bitrate, or the other way around. If you re-encoded as true variable bitrate though, you'd be fairly unlikely to hit the same average bitrate again.

Mind you I'm not sure what encoding modes Handbrake offers, but in theory at least....