Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[6233638]

This is another very good and usefull addition in MC, and still under development I hope. During the development please make sure Stereo Playback from Video is taken care off. To me it seems that the impact on Stereo Video playback is huge. If I play 24/96 Audio files from CD I have an Audiophile performance, when playing this kind of format from Music Video it's not (dynamics are gone) Playing around with Adaptive Volume does not do much good. So my request, please make sure that Stereo Streams from Video are handled without tweaking, I would like it pure, Audiophile....

Check the Dynamic Range (DR) and Dynamic Range (R128) values for both files.

What you are probably hearing, is exactly what R128 was designed to do. R128 plays back both files at the same volume level, regardless of their dynamic range.
Previously, tracks with a compressed dynamic range (a lot of modern music) would sound a lot louder than well mastered tracks with a lot of dynamic range.

Subjectively, most people will pick the louder of two tracks as sounding "better" regardless of how it was mastered.
When you eliminate the difference in volume between the two, you can now identify which track is more dynamic.

So the move towards R128 in broadcast now discourages the use of compressors when mastering music, because all it does is make your music sound worse, rather than making a trade-off between dynamics and loudness. (because you don't get the loudness benefit from it any more)


And if you just want to play videos at full volume with volume leveling disabled, you can set up a separate zone for them using Zone Switch, and disable volume leveling for it.

[Adhara]

Hi,

Is this feature can be used (understand: compatible with) with a room correction system placed after my MC computer ?
When I says "room correction system", I'm thinking about Trinnov or Audyssey.

Is there any link between adaptative volume and the audio analysis feature (R128) ? Is the same "thing" ?
I can not tell the difference between both terms.

Regards.

[6233638]

Is this feature can be used (understand: compatible with) with a room correction system placed after my MC computer ?
When I says "room correction system", I'm thinking about Trinnov or Audyssey.

I don't see any reason it would not work.

Is there any link between adaptative volume and the audio analysis feature (R128) ? Is the same "thing" ?
I can not tell the difference between both terms.

Volume Leveling will adjust the volume of analyzed files during playback so that the average volume is the same. (with a fixed target of -23 LUFS average volume)
If you have a mixed playlist with tracks from lots of different albums, it will mean that the volume should not change between the tracks as it would without volume leveling.
Files from the same album will have one adjustment for all tracks. (to keep intended differences in volume between those tracks)


Adaptive Volume tries to make playback as loud as possible. If volume leveling is not enabled, tracks will be adjusted individually to be as loud as possible, so there may be large variances in volume between each track.
If your files are analyzed, adaptive volume can be combined with volume leveling. This means that all tracks in the current playlist are equal in volume, but the playlist itself is made as loud as possible. (while still keeping the tracks at the same volume)

[Adhara]

I don't see any reason it would not work.
Volume Leveling will adjust the volume of analyzed files during playback so that the average volume is the same. (with a fixed target of -23 LUFS average volume)
If you have a mixed playlist with tracks from lots of different albums, it will mean that the volume should not change between the tracks as it would without volume leveling.
Files from the same album will have one adjustment for all tracks. (to keep intended differences in volume between those tracks)


Adaptive Volume tries to make playback as loud as possible. If volume leveling is not enabled, tracks will be adjusted individually to be as loud as possible, so there may be large variances in volume between each track.
If your files are analyzed, adaptive volume can be combined with volume leveling. This means that all tracks in the current playlist are equal in volume, but the playlist itself is made as loud as possible. (while still keeping the tracks at the same volume)

Thanks for letting me know. Will try to play with this.

[Hendrik]

So my request, please make sure that Stereo Streams from Video are handled without tweaking, I would like it pure, Audiophile....

All these features are optional. If you want it "pure", then just turn it off, and you get the audio as it was encoded in the file.

[Adhara]

If volume leveling only or volume leveling + adaptative volume are enabled, is MC internal volume disabled ?
So that means the only way to play louder or quiet is to deal with DAC volume.

I have some difficulties to understand in wich case I have to use this way or this way and what is the better way...

[DoubtingThomas]

Just finished the re-analysis of my entire library.  Went faster than I expected.

I do have about 15 files that will not analyze.  They just stay at 0% forever and while MC still plays, it's basically locked up and I have to kill it.

Does anyone at JRiver want a copy of one of these audio files to test with?

[Matt]

What you are probably hearing, is exactly what R128 was designed to do. R128 plays back both files at the same volume level, regardless of their dynamic range.
Previously, tracks with a compressed dynamic range (a lot of modern music) would sound a lot louder than well mastered tracks with a lot of dynamic range.

Subjectively, most people will pick the louder of two tracks as sounding "better" regardless of how it was mastered.
When you eliminate the difference in volume between the two, you can now identify which track is more dynamic.

So the move towards R128 in broadcast now discourages the use of compressors when mastering music, because all it does is make your music sound worse, rather than making a trade-off between dynamics and loudness. (because you don't get the loudness benefit from it any more)

This is very well put.

The same goes for moving to R128 in software like JRiver (and hopefully all others eventually).  It's a way to fight the loudness war.

[Adhara]

Hi,

I tried both volume adaptive + volume leveling after having analyzed all my audio files (9k in 2 hours).
But I'm not able to hear a difference with and without. And some audio tracks play louder than others...So I suspect it doesn't work at all for me.

Any explanation ? How to test or verify ?

Thanks

[InflatableMouse]

Just finished the re-analysis of my entire library.  Went faster than I expected.

I do have about 15 files that will not analyze.  They just stay at 0% forever and while MC still plays, it's basically locked up and I have to kill it.

Does anyone at JRiver want a copy of one of these audio files to test with?

Can you download AudioTester and test the files? No install, single executable. Just start it and drag/drop the files onto it.

[InflatableMouse]

Hi,

I tried both volume adaptive + volume leveling after having analyzed all my audio files (9k in 2 hours).
But I'm not able to hear a difference with and without. And some audio tracks play louder than others...So I suspect it doesn't work at all for me.

Any explanation ? How to test or verify ?

Thanks

Few checks you can do.

Check the tags for R128 to see if they have values.
Add a few files to playing now, start playing them and check audio path; hover the mouse over DSP button and see the popup, it shows input, changes and output. Under changes it should tell you volume leveling is used and by how much.
Open DSP Studio and while still playing the songs and disable/enable the volume leveling/adaptive volume options. Unless the changes from volume leveing are 0dB, you should here a difference I would think.

Obviously, DSP studio doesn't work when you're bitstreaming. In this case audio path is empty I think.

[wayneoh]

When I ripped my CDs to the wav and flac using dBpoweramp, I checked the boxes to write track gain and write album gain on the DSP ReplayGain dialog box, but I did not check the box about R128. (I attached a screen cap that I hope shows up.)  Anywho, given that JRiver v19 is now using the R128 standard to analyze loudness and dynamic range of content, does that mean that I should re-rip my CDs with the “EBU R 128 Calculated Gain” box checked in dbpoweramp? And if, god forbid, the answer to that question is yes, then should the “LUFS Target Volume” setting be left at -23 or changed to some other value?

[6233638]

When I ripped my CDs to the wav and flac using dBpoweramp, I checked the boxes to write track gain and write album gain on the DSP ReplayGain dialog box, but I did not check the box about R128. (I attached a screen cap that I hope shows up.)  Anywho, given that JRiver v19 is now using the R128 standard to analyze loudness and dynamic range of content, does that mean that I should re-rip my CDs with the “EBU R 128 Calculated Gain” box checked in dbpoweramp? And if, god forbid, the answer to that question is yes, then should the “LUFS Target Volume” setting be left at -23 or changed to some other value?

You just need to analyze the files with Media Center. Select any files that you want analyzed, right-click them and select Library Tools > Analyze Audio. Disable the "skip analyzed files" option if necessary. (I'm not sure if files created with dBpoweramp will require this option to be disabled)

[Denti]

I have 80k+ songs. Will I need to re analyze all of these every time I add new music?

[JimH]

No.

[jrdiandrea]

I just installed mc19 a few days ago and re-analyzed all of my audio files, about 27k. I now have 2 sets of peak level, volume level & dynamic range results when i look at individual file tags. Is this normal. If not what should I do?!

Thx!

[6233638]

Im just installed mc19 a few days ago and re-analized all of my audio files, about 27k. I now have 2 sets of peak level, volume level & dynamic range results when i look at individual file tags. Is this normal. If not what should I do?!

This is normal. Media Center uses the Volume Level (R128) tags during playback.
The Volume Level (ReplayGain) tags are for compatibility with other players - Media Center can read legacy ReplayGain tags (which are not as good as R128) and after analysis will write ReplayGain 2.0 tags with the R128 data.

Dynamic Range (R128) is a measure of the track's overall dynamic range. (difference between loudest and quietest parts)
Dynamic Range (DR) is a measure of crest factor and produces results compatible with the Dynamic Range Database.

Tracks with higher dynamic range values are typically better mastered than those with lower dynamic range values.

[jrdiandrea]

Makes sense now. Thx!!

[Denti]

No.

But I do need to analyze the new tracks, right?

[6233638]

But I do need to analyze the new tracks, right?

You will. Auto-import can be set to automatically analyze new files though - but it may skip over files if they have existing ReplayGain data. (I'm not sure if that applies to MC19 now due to the move to R128)

[pluto]

Would someone please be so kind as to clarify the meaning of [and distinction between]:

"Write tags when analyzing audio..." (options / library & folders / configure auto-import)

&

"Update tags when file info changes" (options / general / importing & tagging)

Is there any dependency between these two settings?

Clarification would be appreciated - thanks.

[Denti]

So it took three days to analyze my files, but it's done. The result? It works, sure, but the overall volume is WAY lower. I mean, I have to turn my system up significantly to get the same volume I have without the leveling. Is this normal?

[6233638]

So it took three days to analyze my files, but it's done. The result? It works, sure, but the overall volume is WAY lower. I mean, I have to turn my system up significantly to get the same volume I have without the leveling. Is this normal?

Yes, audio is normalized to a target of -23 LUFS. (dB)
If you did not reduce the volume this much, you would have a problem with the volume level changing, as very dynamic tracks would hit the clipping protection.

[Vocalpoint]

Does MC19 still support ReplayGain tags?

I do not believe we got a clear answer on this question? I am getting ready to install a build of v19 soon - but risk any damage to my library (the rest of household will be on v18 for some time yet) if v19 does something to existing RG tags.

So..

1. Does v19 do anything to existing RG tags in files analyzed via v18?
2. Can I install v19 and continue to use all RG data as if the R128 goodies do not exist?

VP

[rjm]

Have been analyzing for 3 days and have another 3 or 4 days to go. Here is an interim report.

Most dvds seem to analyze ok but I had a couple for which the status ran up to something like 200,000% before it completed successfully and changed to Done.

Am collecting a variety of audio and video files that error out when analyzed. Expect 50 or so out of 180,000 so not a problem that I care gets fixed. Does JRiver wish to chase these oddballs down or shall I drop the issue?

I did come across one mp3 album that consistently causes MC to crash during analysis. We should definitely chase this down but I will wait to see if there are any more like this before providing details.

[DoubtingThomas]

I have two versions of Dire Straits / Brothers In Arms.

One version is the 2013 MFSL remaster and the other 24 bit 96khz.

Using "One World" as an example.

Dynamic Range
MFSL: 17
24bit96kHz: 11

The 24bit96kHz definitely sounds "louder" with volume leveling active (adaptive volume: OFF)

I don't have MC18 installed any longer so I can't compare to the old ReplyGain volume leveling.

It's not OMG louder, but it's noticeable when switching back and forth between tracks.

Is this normal for "Volume Leveling" when the DR is different?

[6233638]

Have been analyzing for 3 days and have another 3 or 4 days to go. Here is an interim report.
Most dvds seem to analyze ok but I had a couple for which the status ran up to something like 200,000% before it completed successfully and changed to Done.
Am collecting a variety of audio and video files that error out when analyzed. Expect 50 or so out of 180,000 so not a problem that I care gets fixed. Does JRiver wish to chase these oddballs down or shall I drop the issue?
I did come across one mp3 album that consistently causes MC to crash during analysis. We should definitely chase this down but I will wait to see if there are any more like this before providing details.

It would probably be good if you could send samples to the JRiver team.

I have two versions of Dire Straits / Brothers In Arms.
One version is the 2013 MFSL remaster and the other 24 bit 96khz.
Using "One World" as an example.

Dynamic Range
MFSL: 17
24bit96kHz: 11

The 24bit96kHz definitely sounds "louder" with volume leveling active (adaptive volume: OFF)

I don't have MC18 installed any longer so I can't compare to the old ReplyGain volume leveling.
It's not OMG louder, but it's noticeable when switching back and forth between tracks.
Is this normal for "Volume Leveling" when the DR is different?

There are a couple of possibilities here.

1. Depending on how the albums are tagged, Media Center may be inadvertently treating both tracks as being from the same album, so they have a fixed correction. To check this, only have both versions of the same track in the "now playing" view, with the Volume Level (R128) information visible. Check that the tracks are using the correct level in the Audio Path window.

In my library, I use the Description field to specify the version of an album, which means that Media Center thinks they're from the same album, because its name is the same:


As you can see, when I add the CD track, the volume level for both is changed to −10 dB, even though the currently playing track should be +0.9 dB.
This results in the much better quality SACD track being played back 10.9 dB quieter than the CD one.


2. It's possible that the difference in dynamic range is large enough, that it's not possible for both tracks to perceptually sound the same. R128 does a really good job—better than ReplayGain ever did—but it may not be possible to level things exactly.

3. It may also be that you have run out of headroom, and require more than −23 LUFS allows for. (this is something that will hopefully be addressed in an update)

In your "now playing" view, add an expression column using:

Code: [Select]

Delimit(if(isempty([Peak Level (R128)]),,formatnumber(math(removecharacters(left([Peak Level (R128)],5),/ /+,0)+RemoveCharacters([Volume Level (R128)],/ LU,0)+1),1)),/ dB,)If this is a positive number for any of the tracks, it means that more headroom is required for proper volume leveling.

[DoubtingThomas]

1. Depending on how the albums are tagged, Media Center may be inadvertently treating both tracks as being from the same album, so they have a fixed correction. To check this, only have both versions of the same track in the "now playing" view, with the Volume Level (R128) information visible. Check that the tracks are using the correct level in the Audio Path window.

Both albums had slightly different album names.

2. It's possible that the difference in dynamic range is large enough, that it's not possible for both tracks to perceptually sound the same. R128 does a really good job—better than ReplayGain ever did—but it may not be possible to level things exactly.

3. It may also be that you have run out of headroom, and require more than −23 LUFS allows for. (this is something that will hopefully be addressed in an update)

In your "now playing" view, add an expression column using:

Code: [Select]

Delimit(if(isempty([Peak Level (R128)]),,formatnumber(math(removecharacters(left([Peak Level (R128)],5),/ /+,0)+RemoveCharacters([Volume Level (R128)],/ LU,0)+1),1)),/ dB,)If this is a positive number for any of the tracks, it means that more headroom is required for proper volume leveling.

None are positive, but one was -2.7 ...  (thanks for the expression)

I suspect as you say, it's just the limited dynamic range of the 24bit96kHz version compared to the MFSL version.

It really sucks that the people who do the "remastering" will limit the dynamic range.  One simply can't know that a "24bit96kHz" album is any better than other versions.  But LOL... I own the MFSL version and will be returning the 24bit96kHz version back to a friend...  it was fun to compare.  If there is a fidelity difference between these versions (other than DR) I can't hear it.

[DoubtingThomas]

Do you have very many real world examples in your library where the Peak Level (R128) doesn't have enough headroom to adjust for Volume Level (R128)? In my library of 6400 audio tracks I found just three and it only made a 1.6 dB or less difference. I also would never play back those tracks without the entire album so it wouldn't matter anyway. The one needing the most Volume Level (R128) measures 15.8 and it has a Peak Level (R128) of -15.1 dBTP. The actual adjustment is 14.1 dB.

Only 200 tracks of mine need a positive volume adjustment (.03%). Out of those 200, there are only 4 tracks that I would play in a mixed playlist.

None of mine are >= 0, but I have some that are -0.1

[dtb300]

3. It may also be that you have run out of headroom, and require more than −23 LUFS allows for. (this is something that will hopefully be addressed in an update)

Is this the same issue that may cause some music to clip?  

I have one song (at least one that I noticed last night) that was hitting 100% when viewing the DSP window in setup.   The only thing I had checked was Up-sampling, so I hit STOP, unchecked up-sampling, started the song again, and it still hit 100% quite often.

Running 19.0.38 with volume set to Internal.

[6233638]

Is this the same issue that may cause some music to clip?  

No, Volume Leveling should not allow music to clip - it simply means that the track will be played quieter than "ideal" for leveling due to clipping protection.

[dtb300]

No, Volume Leveling should not allow music to clip - it simply means that the track will be played quieter than "ideal" for leveling due to clipping protection.

I thought I also tried it with Volume Leveling, but will double-check that tonight to be sure.  I report back with my findings....now I need to remember the song - LOL

[sillibil]

Mr.C. "The analysis data is stored in the library, and by default in the new tags (and some compatibility tags which are auto-generated).  These are stored in file tags, so an Update Library (from tags) would re-read them:"

I got a question, I'm backing up my files extern about 2.5 Gb. I did analyze my music files , some backup programs with incremental sync.
don't write the DR 128 in my backup files, like Microsoft Richcopy. Does the quote means that in case of HDD failure and restore my library I can Update my library from restore library all my DR 128 taggs are there again or not? Or should make a complete new backup.

[Matt]

I got a question, I'm backing up my files extern about 2.5 Gb. I did analyze my music files , some backup programs with incremental sync.
don't write the DR 128 in my backup files, like Microsoft Richcopy. Does the quote means that in case of HDD failure and restore my library I can Update my library from restore library all my DR 128 taggs are there again or not? Or should make a complete new backup.

The library or a library backup contains all of your data (except the actual media files and cover art).  The file's tags are essentially just a backup and also useful if you move files to another program.

In other words, if you have the library, you have the audio analysis data and can always apply it to the tags at a later date.

[sillibil]

Thanks Matt, that's wonderful and a great future.

[Arcturus]

Also new to this and have 2 questions.

1. Will analyzing my audio change the files in any way or is the data stored outside of the file. Just wondering because I have my music sync'd to a drive on a server to keep it safe. Don't mind if its going to have to resync the whole thing again.  

2. How will this effect mixed albums that come with a .cue sheet. Example I have an album that's 2 hours long but in a single .FLAC but with a .cue file to break up each track in JRiver. So how would that work to normalize the album?

EDIT. Decided to take the plunge and let it plow through the 28,000 audio files.
I am curious though does it have to read the entire file to do what its doing or how exactly does that work. Same for video is it going to have to read a 4-16gb file just to do its job?

Also notice it added and or modified the BPM field of the id3 tags so never mind the first question.

[Matt]

1. Will analyzing my audio change the files in any way or is the data stored outside of the file. Just wondering because I have my music sync'd to a drive on a server to keep it safe. Don't mind if its going to have to resync the whole thing again.

By default the audio analysis results are stored in the library AND file tags.  If this bothers you, you can turn off writing of any field to the tag using Options > Library & Folders > Manage Library Fields.  I would not recommend doing this.

I am curious though does it have to read the entire file to do what its doing or how exactly does that work. Same for video is it going to have to read a 4-16gb file just to do its job?

It has to read and decode the entire file to analyze all the audio.  It's like playing, only at full speed instead of 1x.

[William-NM]

Just wanted to say thank you for these features. I notice that I no longer need to adjust the volume, even on playlists that include files like old 64kb mono tunes from early jazz 78's. Amazing! I re-analyzed my 20K songs the other day, and was listening with the old Sennheiser HD-580's last night, and I'm convinced that even the mp3's sound more transparent, crisp and defined than they used to. 

Will this hold true when playing to a tab via gizmo? My Asus died yesterday, so I'm shopping for another (don't really mind  ), and I stream my workout mix back to the 'gym' - have noticed major vol. fluctuations in the past. Thanks, again.

[6233638]

I'm convinced that even the mp3's sound more transparent, crisp and defined than they used to.

They may very well do, as Volume Leveling should help avoid inter-sample clipping, which is common with lossy formats.

[HiFiTubes]

I have been noticing quite a reduced volume in .dff files and I have 4 copies of that Jones album so I wonder if it is an issue? Doesn't DSD have -6db or something? But these ISO are being converted to DXD on fly by MC. I'll have to look at the tags tonight.

Anyway, can someone comment on best settings for using RS128 with a DAC with digital volume control? If I want to do most volume adjustment in PC, and minimize use in the DAC (Auralic Vega). What is Volume Disabled vs. Internal Volume with regard to using RS128?

thanks!

[realysm42]

Guys,

Thanks for rthe feature.

Does enabling just levelling hurt dynamics and sq?

If yes, is there anything that can be done about it?

If I disable I internal volume, so its always @100%, does this feature still work? I ask because I'm getting an analogue preamp installed and would like to us purely that.

Thanks for your help.

[6233638]

Does enabling just levelling hurt dynamics and sq?

As long as your audio device is at least 24-bit, it should not hurt sound quality.

If anything, using volume leveling should improve your perception of dynamics.

If I disable I internal volume, so its always @100%, does this feature still work? I ask because I'm getting an analogue preamp installed and would like to us purely that.

Yes, Volume Leveling will still work.

[MikeO]

I previously asked on a different thread if volume levelling will affect quality. I should have done the test , being lazy !

I did the test , I used 2 of my "test tracks" , Joni Mitchell - Blue - A Case of You , and Dire Straits - Love Over Gold - Private Investigation both FLAC via MC from remastered CD's

Out via ASIO to  a  Cambridge Audio DAC Magic , then Denon AV amp (set Direct) and HD800 .

Even to my aging ears , I believe I can tell a difference on the tops , sibilance as less pronounced and the aggressive "twang" of the guitar in Dire Straits was definitely missing.

I saw a dramatic reduction in volume of some tracks , presumably because I have a very high level track or two in the library

Are there settings to adjust ?. I have switched it off for now

[mwillems]

I previously asked on a different thread if volume levelling will affect quality. I should have done the test , being lazy !

I did the test , I used 2 of my "test tracks" , Joni Mitchell - Blue - A Case of You , and Dire Straits - Love Over Gold - Private Investigation both FLAC via MC from remastered CD's

Out via ASIO to  a  Cambridge Audio DAC Magic , then Denon AV amp (set Direct) and HD800 .

Even to my aging ears , I believe I can tell a difference on the tops , sibilance as less pronounced and the aggressive "twang" of the guitar in Dire Straits was definitely missing.

I saw a dramatic reduction in volume of some tracks , presumably because I have a very high level track or two in the library

Are there settings to adjust ?. I have switched it off for now

Volume leveling works by (mostly) reducing the volume of tracks to recenter them to a specific standard level.  The result is that (with leveling on) most music will sound about the same volume regardless of compression (i.e. heavily compressed music will not "sound louder" due to compression).  Generally speaking, louder music (even slightly louder music) sounds "better" to most people, which is how the loudness wars got started to begin with.  

You don't mention whether you readjusted the volume in your experiment so that you were listening at the same "effective" volume level in both cases, but, if you haven't, I would recommend doing that before coming to any firm conclusions about volume leveling.

[6233638]

Even to my aging ears , I believe I can tell a difference on the tops , sibilance as less pronounced and the aggressive "twang" of the guitar in Dire Straits was definitely missing.

If you add an expression column with the following code, are there any values above 0?

Code: (True Peak Level) [Select]

math(removecharacters(listitem([Peak Level (R128)],0),+/ dBTP,0)+1)
If they are above 0, than means what you're probably hearing is the removal of inter-sample clipping, which is a very bad artifact of modern mastering processes.
If not, you're just hearing a difference in volume.

I saw a dramatic reduction in volume of some tracks , presumably because I have a very high level track or two in the library

Everything is normalized to -23 dB, so it depends how loud the track was mastered.

[MikeO]

Hi

While not "metering" the comparison to ensure similar volume levels in my head phones , I did the adjustment so to my ears they both sounded the same volume , there was not a massive difference

I created the Expression with the maths you provided and is see only zero values for the > 30,000 tracks in my library , is that to be expected ?

Mike

[6233638]

I created the Expression with the maths you provided and is see only zero values for the > 30,000 tracks in my library , is that to be expected ?

Have you analyzed your library in MC19 yet?

[Adhara]

Hi,

What is the behavior when there are several audio tracks (languages) on a video ?
Which one is the reference ? Or maybe they are all analized ?

Regards.

[mojave]

Hi,

What is the behavior when there are several audio tracks (languages) on a video ?
Which one is the reference ? Or maybe they are all analized ?

Regards.

The one that is analyzed is the one that is played when you select the movie. You can change the audio track during playback, then analyze after playback has stopped. The last used audio track will be analyzed.

If you want/need several tracks analyzed, the create a Particle of the movie and select a different track for the Particle.

[Matt]

What is the behavior when there are several audio tracks (languages) on a video ?

The analyzer will analyze the current audio track (the one you would hear if you played the file).

You can configure how default audio tracks are selected in Options > Video.

Or you can play a video and change the audio track so that the current audio track is changed (for analysis, serving to DLNA, Gizmo, etc.).

(I cross-posted with mojave.  As usual, he's faster than me.)