INTERACT FORUM
More => Old Versions => JRiver Media Center 19 for Windows => Topic started by: DoubtingThomas on July 18, 2014, 11:01:42 am
-
While I can't go back and test... I believe the old "replay gain" was more effective than the current "volume leveling" in v19.
For example... The Muddy Waters / Folk Singer (24bit96Khz) album is VERY loud compared to most other albums.
I have re-analyzed this album and none of the values changed.
-
While I can't go back and test... I believe the old "replay gain" was more effective than the current "volume leveling" in v19.
It is not. Volume leveling is based on Replay Gain.
-
Okay, but either way that Muddy Waters album I mentioned is very LOUD compared to other albums, and there are other albums that are just too loud.
Would you like me to dropbox one of the tracks on the album for you?
-
Okay, but either way that Muddy Waters album I mentioned is very LOUD compared to other albums, and there are other albums that are just too loud.
Would you like me to dropbox one of the tracks on the album for you?
Sure. matt at jriver dot com.
-
I've sent you the dropbox link, thanks.
-
I got the file.
However, it plays at about the same volume as other analyzed files.
Advice?
-
While I can't go back and test... I believe the old "replay gain" was more effective than the current "volume leveling" in v19.
For example... The Muddy Waters / Folk Singer (24bit96Khz) album is VERY loud compared to most other albums.
I have re-analyzed this album and none of the values changed.
Enable track-based leveling instead of album-based.
-
It is not. Volume leveling is based on Replay Gain.
Matt, are you saying that Volume Leveling is currently using the Replay Gain figure instead of R128? If that's not what you're saying, what did you mean? Your short one-liners are often difficult to understand what you're really trying to say.
I have to agree with the OP, for whatever reason there can still be quite a variance between tracks. Muddy Waters is a good example because it is mixed rather bright (his vocal mostly) and is very annoying to listen to at your normal reference level.
Last week I was experimenting with this trying to find some indication that represented level adjustments that I hear from track to track. By far, the old Replay Gain figure was closer to reality, but either group could still be off track to track by over 6db. My reference listening level is between 83db and 86db, a comfortable level on most well recorded music when played on a well balanced system. But throw Muddy in there, and you *have* to turn it down.
The only way I have found to mitigate this is to listen at lower levels, like around 70db or less where your ear is less sensitive to level changes. But that takes some of the fun out of it!
And forget Peak Normalize for anything, as it frequently allows the signal over 0db reference level and clipping. The only way to get extremely clean audio on everything you play is with Volume Leveling ON, Adaptive Volume OFF, and Clip Protection OFF. Even at that you'll see peak output levels hovering around -6 to -12db, but with some songs pushing up to around -1db at the output of the DSP studio.
--Bill
-
Matt, are you saying that Volume Leveling is currently using the Replay Gain figure instead of R128? If that's not what you're saying, what did you mean? Your short one-liners are often difficult to understand what you're really trying to say.
Volume Leveling uses Replay Gain.
-
Volume Leveling uses EBU R128 since MC19. The algorithm is an industry standard and more reliable than replay gain.
It'll compute the Replay Gain value in the database based on the R128 figures (basically a offset of a few dB)
-
Enable track-based leveling instead of album-based.
That would be very bad since I play albums and I would not want the volume jumping up/down when playing for example Pink Floyd.
-
First thing, make sure that you only have Volume Leveling enabled, and that Adaptive Volume is disabled.
Secondly, I'm not entirely clear - is your issue that tracks on the same album have large variances in volume, or that album-to-album is having large changes in volume?
The new R128-based leveling system in MC19 should produce better results than the old ReplayGain system ever did.
The ReplayGain v2 spec is moving to an R128-based leveling system as well, as they consider it to produce better results too.
-
First thing, make sure that you only have Volume Leveling enabled, and that Adaptive Volume is disabled.
Secondly, I'm not entirely clear - is your issue that tracks on the same album have large variances in volume, or that album-to-album is having large changes in volume?
The new R128-based leveling system in MC19 should produce better results than the old ReplayGain system ever did.
The ReplayGain v2 spec is moving to an R128-based leveling system as well, as they consider it to produce better results too.
Adaptive Volume is disabled.
I sent Matt a track that plays loud to me, and he says it plays the same volume as other tracks... strange.
I'll try to gather some better info for future reports.
-
Are you sure the file is analyzed? What value does the field "Volume Level (R128)" have?
-
Are you sure the file is analyzed? What value does the field "Volume Level (R128)" have?
The whole album is loud compared to other albums.
(https://dl.dropboxusercontent.com/u/75405336/MuddyWaters.jpg)
-
The whole album is loud compared to other albums.
This album is VERY dynamic compared to other albums but I have both the Mobile Fidelity release and the 24/96 - both analyzed and they plays perfectly against anything else in the library. No jarring volume jumps or anything out of the ordinary.
What settings - exactly - do you have on in DSP studio for playback? Volume leveling should be the only thing on - in addition to Output format of course.
VP
-
This album is VERY dynamic compared to other albums but I have both the Mobile Fidelity release and the 24/96 - both analyzed and they plays perfectly against anything else in the library. No jarring volume jumps or anything out of the ordinary.
What settings - exactly - do you have on in DSP studio for playback? Volume leveling should be the only thing on - in addition to Output format of course.
VP
I have the Equalizer checked too... so AutoEQ can use it. I have no eq adjustments for this album.
-
James Taylor / Sweet Baby James (24bit192khz) plays loud too.
-
James Taylor / Sweet Baby James (24bit192khz) plays loud too.
I have this release as well..and it plays perfectly - especially with the fact it's a quiet record to begin with.
But I do not have anything else inline except Volume Leveling. Still not sure what AutoEQ is or why you need to have EQ in your chain at all...
Clearly this is a case of your setup doing something to make certain releases sound louder. All I can think of is either you are not analyzing certain items correctly (or at all) OR that EQ has to be doing something to your levels.
Over here - Volume Leveling in MC is on 100% of the time on every release and - and if I heard anything out of the ordinary - I would certainly be the first to say so.
VP
-
I have no explanation for it so far.. but so far all the loud track are high sampling bit rates.
For 99% of my tracks the Equalizer is set flat.
I only have Volume Leveling turned on and all tracks are analyzed.
-
I have no explanation for it so far.. but so far all the loud track are high sampling bit rates.
Well - as I mentioned - I have a wide selection of 24/192, 24/96 and so on - none of my "high sample rate" files exhibit any sort of issue.
Next step for you is to take the EQ out of the chain completely - and reanalyze the James Taylor and then try it with just Volume Leveling on by it's lonesome.
Q: What do you have set in Output Format for when a 24/192 file is encountered?
Mine is set to None in Output Encoding and No Change for a 192,000 hz file. Over in Channels - it's set to 2 channel (Stereo) and Mixing is set to JRSS Mixing (recommended)
Update when you can.
VP
-
What is JRSS Mixing?
I'm running a digital coax from my sound card to my audio system processor.
(https://dl.dropboxusercontent.com/u/75405336/Clipboard.jpg)
-
I see that you have Clipping Protection disabled - that's a very bad idea.
And if you have had to enable it, it suggests that your EQ settings might be too high. Try reducing the pre amp setting. (if your highest value is +6 for example, set pre-amp to -6)
You should also enable JRSS Mixing on the right.
-
I see that you have Clipping Protection disabled - that's a very bad idea.
And if you have had to enable it, it suggests that your EQ settings might be too high. Try reducing the pre amp setting. (if your highest value is +6, set pre-amp to -6)
You should also enable JRSS Mixing on the right.
I never ever use the EQ to raise a level, only to lower the bass for badly mixed tracks.
What is JRSS Mixing...
-
What is JRSS Mixing...
If your source content's number of channels is different from your output settings (in this case stereo) it will mix the channels (in the case of mono sources, upmixing; multichannel, downmixing) so that your audio sounds as good as possible over your output # of channels.
I don't know the technical details of this process but JRSS is supposed to be one of the best mixers.
http://wiki.jriver.com/index.php/Mixing
-
If your source content's number of channels is different from your output settings (in this case stereo) it will mix the channels (in the case of mono sources, upmixing; multichannel, downmixing) so that your audio sounds as good as possible over your output # of channels.
I don't know the technical details of this process but JRSS is supposed to be one of the best mixers.
http://wiki.jriver.com/index.php/Mixing
Thanks for link. I only play stereo audio using MediaCenter. I only use MediaCenter for audio.
-
Thanks for link. I only play stereo audio using MediaCenter. I only use MediaCenter for audio.
Turning on JRSS won't hurt anything then. Basically, the difference is this:
With it configured with JRSS Off (set to No Upmixing or Downmixing) then if you ever try to play a file that does not have the exact number of channels you have configured, playback will FAIL and you'll get an error message. So, if you try to play a multichannel track, or a mono track, it'll pop up an error message and make you sad.
With JRSS On, then those sources will be automatically mixed using the best possible mixing algorithm, and those that DO match (the vast majority, or all of your sources currently) will be left untouched.
That's the answer.
-
Turning on JRSS won't hurt anything then. Basically, the difference is this:
With it configured with JRSS Off (set to No Upmixing or Downmixing) then if you ever try to play a file that does not have the exact number of channels you have configured, playback will FAIL and you'll get an error message. So, if you try to play a multichannel track, or a mono track, it'll pop up an error message and make you sad.
With JRSS On, then those sources will be automatically mixed using the best possible mixing algorithm, and those that DO match (the vast majority, or all of your sources currently) will be left untouched.
That's the answer.
LOL... well I don't want to be sad.... so I've turned JRSS on... thanks.
-
I split the posts about sample rates and exclusive access (http://yabb.jriver.com/interact/index.php?topic=90541.0).
-
Just wanted to throw in this graphical representation of MC's volume levelling at work..
(https://dl.dropboxusercontent.com/u/7082504/Media%20Center/r218.JPG)
This is 20 minutes of random tracks. You can see the music is correctly normalized around the -23 (integrated (LUFS)) point as specified in R128.
You see some variances around that point depending on the dynamics in the track but generally speaking it's spot on.
Not saying the tracks being discussed cannot cause issues, but just wanted to throw this image in as confirmation that things are generally running as expected.
-
Okay, for anyone with these tracks.
Sinead O'Connor / Lion and The Cobra / Troy
Sinead O'Connor / I Do Not Want What I Have Not Got (reissue) / Troy (live)
The studio version is significantly lower in perceived volume than the live version.
Both are analyzed, and I re-analyzed them both today, no change from prior values.
(https://dl.dropboxusercontent.com/u/75405336/Clipboard.jpg)
I've sent Matt dropbox links to both audio tracks.
-
Unless there is a bug in the analysis code, I'm not sure there is much that can be done.
R128 is not a standard developed by JRiver, it was created by the ITU (https://en.wikipedia.org/wiki/International_Telecommunication_Union)/EBU (https://en.wikipedia.org/wiki/European_Broadcasting_Union).
Without having access to the tracks, it's difficult to say why they sound different.
Looking at your screenshot, there is a not-insignificant difference in dynamic range between the two.
-
Geoff Achison / BamBoo Room Blues is also VERY LOUD compared to other albums.
-
Okay, for anyone with these tracks.
Sinead O'Connor / Lion and The Cobra / Troy
Sinead O'Connor / I Do Not Want What I Have Not Got (reissue) / Troy (live)
I will check these out later this evening. I have both here.
Question: In your screen cap - which is which. Which one has a DR of 6 or 9?
VP
-
Geoff Achison / BamBoo Room Blues is also VERY LOUD compared to other albums.
I can't even find a reference to this record.....
?
VP
-
I will check these out later this evening. I have both here.
Question: In your screen cap - which is which. Which one has a DR of 6 or 9?
VP
The studio version is 6 and the live version is 9.
-
The studio version is 6 and the live version is 9.
Thanks!
VP
-
Okay, for anyone with these tracks.
Sinead O'Connor / Lion and The Cobra / Troy
Sinead O'Connor / I Do Not Want What I Have Not Got (reissue) / Troy (live)
The studio version is significantly lower in perceived volume than the live version.
Both are analyzed, and I re-analyzed them both today, no change from prior values.
I took a good listen to both of these and I am not sure it's a fair compare. The studio version is very quiet until about 4:30ish or so when it really ramps up and takes on the DR6 the analysis says it has. Because of it's makeup (quiet for 3/4 and loud for the other quarter) it's not like MC can "ride" the volume on this and completely level it out at all stages as it plays.
The live version is a DR9 but it's constant throughout volume wise - giving it a uniform vibe throughout. These two tracks will (and should) have a wide swing due to how they are recorded.
That said - not sure exactly what you are expecting MC to do here - as these kinds of things will happen. Pink Floyd is a great example where things get very very quiet in some spots (The Wall has many tracks like this) - where the song is barely there and then suddenly the band comes flying in. But MC will not (and should not) be attempting to "level" an entire piece - compensating for the nearly silent passages where it's sound effects etc and make that as loud as when the band is grinding away.
No one said Volume Levelling was perfect but there will be times when not much can be done - and that's not the fault of MC. In this case - it's the makeup of that first track that makes it look like the volume variance is huge - cause it is :)
But I can think of many segues that I could put together where the same would be true.
Cheers,
VP
-
I took a good listen to both of these and I am not sure it's a fair compare. The studio version is very quiet until about 4:30ish or so when it really ramps up and takes on the DR6 the analysis says it has. Because of it's makeup (quiet for 3/4 and loud for the other quarter) it's not like MC can "ride" the volume on this and completely level it out at all stages as it plays.
Dynamic range compression (Adaptive Volume in Night Mode) may help with this sort of thing.
But it goes against the goals of leveling and is reducing the dynamic range of your good tracks to match the bad ones.
-
In my memory.... I never had to manually adjust the volume between tracks using the old ReplayGain.
Now there are just some tracks that are simply too loud using Volume Leveling that blast me out of the room and I'm forced to manually adjust the volume.
None of my Pink Floyd albums require any manual volume adjustment.
When I have PlayingNow set with like 10 days of music, and when playing back some of those tracks, the volume is way too high, then I notice. I never noticed this with the old ReplyGain.
-
In my memory.... I never had to manually adjust the volume between tracks using the old ReplayGain.
I did. Not often - but it happened. I have had almost no issues whatsoever with the new R128 spec in MC. Much better in my opinion that ReplayGain ever was (or could be)
Now there are just some tracks that are simply too loud using Volume Leveling that blast me out of the room and I'm forced to manually adjust the volume.
Again - Volume Levelling was never touted to handle 100% of all tracks ever recorded without exception. You will hit some stuff (as you have seen) that simply cannot be levelled to your expectations. This does not mean there is anything wrong with MC. It just means that under "current implementation" - your track cannot be handled to your satisfaction.
That Sinead track is a prime example - but I did take that live version of Troy and stuck it into a random mix - I did not need to pull down the volume when it came on nor did I notice it sticking out whatsoever. It all comes down to what's in front of it and what's coming next that will give the appearance of a wide volume swing.
I maintain that the very mature of the track and it's makeup will determine whether or not the track "appears" to be too loud (or too soft) - not that MC (or any other software) is making a mistake. MC calculates all it's levelling parameters based on what it "sees" in the makeup of the track content.
But no software that I know could make every possible track every recorded "appear" level with everything else out there.
I also noticed this - that you use "album based" leveling?:
That would be very bad since I play albums and I would not want the volume jumping up/down when playing for example Pink Floyd.
If that is on but you are playing track based playlists - you will see differences as well.
I play albums WAY more than playlists and I use track based levelking on everything and have never heard a problem on any album playback. My Pink Floyd albums play back perfectly with NO album based leveling on whatsoever.
VP
-
Enable track-based leveling instead of album-based.
How/where does one even set this?
VP
-
How/where does one even set this?
VP
That would totally ruin Pink Floyd.
-
That would totally ruin Pink Floyd.
Well - until someone can point me in the direction of how enable "track" or "album" based leveling - I will assume I am using "track" right now.
Every PF album here plays perfectly - just as I have always known them to.
VP
-
There is no such option anymore, you're always using Album-based leveling when playing normal music.
Track based is used for Podcasts, Videos and things like that where a volume relation between individual tracks is not given.
Album-based leveling ensures that intentional volume differences within one album are preserved, like the artist may have intended.
-
There is no such option anymore, you're always using Album-based leveling when playing normal music.
Track based is used for Podcasts, Videos and things like that where a volume relation between individual tracks is not given. Album-based leveling ensures that intentional volume differences within one album are preserved, like the artist may have intended.
Thanks!
OK - now that that is solved...
@Doubting Thomas
I guess you will have to accept that fact that you will encounter tracks - due to to their makeup and how MC assigns it's analysis parameters - that simply sound louder than other things in a playlist. I have tested many of the examples that you posted including the Muddy Waters etc - and they all play fine here.
With respect to MC's implementation of the R128 spec and volume leveling - I think it's superb and have not encountered any issues so far within the 65000 song library I preside over. But if I do - I will be sure to revisit this thread.
Cheers,
VP
-
@Doubting Thomas
I guess you will have to accept that fact that you will encounter tracks - due to to their makeup and how MC assigns it's analysis parameters - that simply sound louder than other things in a playlist. I have tested many of the examples that you posted including the Muddy Waters etc - and they all play fine here.
Since I'm currently using AutoEQ... I can lower the "Preamp" slider when I encounter loud tracks.
I'm not good at accepting things....
-
Since I'm currently using AutoEQ... I can lower the "Preamp" slider when I encounter loud tracks. I'm not good at accepting things....
Well - you are insisting on having that EQ inline - which I do not (and will not) - and that alone (despite your insistence that it is not contributing to anything) could be the difference between why you and I are hearing different things.
I can state with 100% accuracy that I have ONLY volume leveling on - and every single track I play - whether full album or within a playlist is perfectly in the pocket and I have never encountered a segue or album playback that required me to go "wow - that's loud" or have to ever alter my volume control.
To wit: The volume knob on my Teac UD-501 actually has a gentle dust film on it (and no fingerprints) from 6 months or more of never touching the knob. That's how I know VL is working fine here.
YMMV
VP