INTERACT FORUM
More => Old Versions => JRiver Media Center 19 for Windows => Topic started by: Matt on July 17, 2013, 11:43:50 am
-
Overview
Adaptive Volume is a system to keep the level of the input signal near full level.
There are several use-cases where this feature is useful:
- You use Volume Leveling and don't want the volume of everything turned down a lot unless it's actually necessary to level the volume of the current playlist
- You would prefer to run your DAC close to full level on any input (more important for DSD and classical where the signal may never approach full scale)
- You watch movies at night or in a situation where explosions are too loud and dialog is too quiet
- You're trying to get enough volume out of a low power system, like during playback on laptop speakers.
MC18 added an adaptive volume mode available for video playback. MC19 promotes this system to be available for any media type, and improves it in several important ways.
This screenshot shows Adaptive Volume in MC19 and explains the modes a little bit:
(http://files.jriver.com/images/2013/adaptive_volume.png)
Peak Level Normalize
This mode requires special consideration. It analyzes the current playlist to find the peak level of the entire playlist by using the values gathered during Audio Analysis. It also considers the peak level of the entire playlist if Volume Leveling is engaged. Then, it normalizes the playlist to this peak level. Since this applies a fixed gain, the blue light will illuminate and Audio Path will say "peak level normalize (fixed)".
If the files being played have not been analyzed using the Analyze Audio tool, the peak level will be learned during playback. In this mode, Audio Path will report "peak level normalize (adaptive)" and the blue light will not illuminate. When used in this mode, the first loud part of a movie or song may cause the volume to be turned down. Once it has turned down, the peak level has effectively been learned and the volume will not be adjusted (unless something even louder happens).
Let's describe a simple example of why this mode might be useful. If you play Adele's 21 album, a modern best-seller mastered in a modern way, and enable Volume Leveling, it will request that all songs be turned down by 14.4dB. This is so that if you play something mastered more conservatively, it will match Adele's volume. However, if you are _only_ playing Adele, you're going to be applying the same -14.4dB to all the songs you hear. You didn't actually need to turn the music down at all to achieve a level volume. Peak Level Normalization will understand this and add the 14.4dB back (internally it's smart enough to do nothing instead of doing two offsetting changes).
Said another way, if you use Volume Leveling and Peak Level Normalization, you will get the loudest playback of the current playlist that maintains equal volume between tracks and prevents all clipping. The volume between tracks in the playlist will be the same, but the volume between different playlists could be different (since each playlist will have a different peak level normalization value).
-
reserved
-
Looks good.
A couple of things though: in the Peak Level Normalize option, I think it reads better as "Boosts low volume content, while preserving dynamic range."
As it is written, it reads as if peak level normalization is required to preserve dynamic range.
As R128 analysis also includes a dynamic range measurement, have you considered fixed levels of dynamic range compression? Even possibly trying to compress dynamic range into user-specified value?
I'm not sure how feasible that would be, or if it's a waste of development time though. I suspect (hope) most people would either use the Peak Level Normalize option, or have adaptive volume disabled altogether.
I still think it's even more important to respect the -1.0 dBTP specified by R128 when using peak normalization for analyzed files, and it should perhaps be set even lower to -3dB with files which have not been analyzed. (assuming adaptive volume is looking at samples and not inter-sample peaks) This AES paper (http://www.tcelectronic.com/media/1018175/lund_2006_stop_counting_samples_aes121.pdf) is the one referenced in Tech 3343 (http://tech.ebu.ch/docs/tech/tech3343.pdf)
For a mixing engineer, advice is simple: If you mix to digital, don’t peak higher than -3 dBFS on a Sample meter.
…
Most consumer CD players and broadcast processors will survive when the signal is lowered by 3 dB, while data reduction codecs (MP3, AAC, DTS, AC3 etc.) may require up to 5 dB attenuation or more, depending on the data rate.
We have this now. ;D
-
Just a grammar correction: . . . You use Volume Leveling and don't want the volume of everything
-
The volume between tracks in the playlist will be the same, but the volume between different playlists could be different (since each playlist will have a different peak level normalization value).
Does all three modes work in this way? If I do not want to turn the volume knob, ever, should I refrain from using this feature altogether?
Also, it is hard to understand from your explanation what happens if all files in the library are analyzed and files are added to or removed from the playlist. Will there be a change in volume when the next track comes on? Are already played tracks in the playlist taken into account if a recalculation is made?
-
Does all three modes work in this way? If I do not want to turn the volume knob, ever, should I refrain from using this feature altogether?
Volume leveling on its own accomplishes this. Volume leveling combined with adaptive volume will level the current playlist, but play it as loud as possible without clipping. (so you have to adjust volume when starting playback, and potentially when adding or removing tracks)
Adaptive volume without volume leveling will adjust the peak of every track to be as loud as possible. This will not sound level at all.
-
Volume leveling on its own accomplishes this. Volume leveling combined with adaptive volume will level the current playlist, but play it as loud as possible without clipping. (so you have to adjust volume when starting playback, and potentially when adding or removing tracks)
Adaptive volume without volume leveling will adjust the peak of every track to be as loud as possible. This will not sound level at all.
This is exactly right with one small clarification.
Adaptive Volume without Volume Leveling will adjust to the peak of the playlist in 'Peak Level Normalize' mode, not each track.
The other modes are more adaptive so it's a blurry line if they're adjusting the track or playlist.
-
Thank you for the answers!
I take it this feature is not for me then, except perhaps when I, out of consideration for my house cohabitants, feel like being in a "night mode". I will refrain from further questions about it.
I am happy with just smart volume leveling.
-
Adaptive Volume without Volume Leveling will adjust to the peak of the playlist in 'Peak Level Normalize' mode, not each track.
So there's no difference whether volume leveling is enabled or disabled in that case? (that's fine, but it seemed like this solved the request from that one guy that wanted the highest SNR at all times)
-
So there's no difference whether volume leveling is enabled or disabled in that case? (that's fine, but it seemed like this solved the request from that one guy that wanted the highest SNR at all times)
It gives that guy the highest peak level _while_ maintaining a constant gain across the playlist. So it's close to what he wanted, but maybe not exactly. He could always play one track at a time ;)
-
Will this keep consecutive tracks from the same album from being 'adjusted'? i.e. if one quiet song is 'supposed to' lead to a loud one, per original mastering, I want to hear it that way. (Dark Side of the Moon)
Is that possible?
-
Will this keep consecutive tracks from the same album from being 'adjusted'? i.e. if one quiet song is 'supposed to' lead to a loud one, per original mastering, I want to hear it that way. (Dark Side of the Moon)
Is that possible?
Volume leveling combined with adaptive volume set to peak normalization should do that.
-
Volume leveling combined with adaptive volume set to peak normalization should do that.
All scenarios will preserve relative album volume differences when the album is played in order:
Volume Leveling
Volume Leveling + Adaptive Volume
Adaptive Volume
-
Volume leveling combined with adaptive volume set to peak normalization should do that.
Cool. I just saw this in another post also. I'm happy :)
- Smarter Volume Leveling that automatically respects intentional between track levels when playing from an album
-
To check I'm reading all this correctly (so If I don't get it the first post from Matt may need some changes). I want:
- When playing the Album I'd want it MC to adjust the volume levelling based on the whole Album and leave any volume difference between tracks in tact, but
- When playing a playlist of a mix of songs from various Albums then I would want volume levelling to apply to all tracks.
To do this I :
1) check Volume Levelling and Adaptive Volume (Peak Level Normalisation) (also is this the default setup?)
2) to work "most effectively" I should also run Analyse Audio over all files to get the new values
Thanks
Nathan
-
Adaptive Volume is only when you want the current playlist to be played as loud as possible. (while avoiding clipping)
If you are using peak normalization and have a playlist where the maximum peak volume is -10dB, it will boost the volume of all tracks in the playlist by +10dB for example.
I personally dislike adaptive volume, because it means that you have to change the volume control each time you start playing something new.
Perhaps this needs to be made clearer somehow.
Volume Leveling (without Adaptive Volume) tries to normalize the volume of all tracks around a fixed level of -23 dB. (technically Loudness Units, or LU, when talking about R128 values)
This means that playback could be considerably quieter, but -23dB gives you enough headroom that it should mean all tracks are able to be normalized to the same value.
With everything normalized to -23 dB, in theory, you would never have to touch Media Center's volume control again.
When you have a playlist of mixed tracks, they are individually adjusted to play at -23dB. (or as close as possible - perceptual analysis is very hard to get completely right)
If you are playing albums rather than a mixed playlist, the average value for the whole album is adjusted to -23dB, rather than adjusting each track individually, so that album dynamics are preserved.
When the two are combined, volume leveling normalizes the volume of all tracks (to -23dB) and then adaptive volume increases the volume of that normalized playlist to be as loud as possible. (e.g. +10dB if the maximum peak level in the playlist is still -10dB)
Files need to be analyzed for either of these features to work well.
-
Another Q regarding the use with Video:
1) Music Video: I plan to Analyse Audio on these as the vol and presume it will work as per my post abovev
2) Movies / TV Eps: I do NOT plan to Analyse Audio on these as they tend to run for a longer period so changing my AV Vol manually on the RC works fine. If these files are NOT Analysed will MC try to adjust vol levels as playback of the movies goes along (using Volume Levelling and Adaptive Volume (Peak Level Normalisation) )? I hope not ;D
-
To make things easier to understand, perhaps "Peak Normalization" should be renamed to something like "Maximize Volume" with a description saying that it tries to play the current playlist as loud as possible without interfering with dynamic range?
And Night Mode/Small Speakers would probably need something that explains they compress the dynamic range to go even louder, but may result in volume fluctuations.
Another Q regarding the use with Video:
1) Music Video: I plan to Analyse Audio on these as the vol and presume it will work as per my post abovev
2) Movies / TV Eps: I do NOT plan to Analyse Audio on these as they tend to run for a longer period so changing my AV Vol manually on the RC works fine. If these files are NOT Analysed will MC try to adjust vol levels as playback of the movies goes along (using Volume Levelling and Adaptive Volume (Peak Level Normalisation) )? I hope not ;D
You'll get the "standard" -10dB for all videos which have not been analyzed if volume leveling is enabled.
Adaptive volume may adjust the volume dynamically during playback if the file has not been analyzed.
-
You'll get the "standard" -10dB for all videos which have not been analyzed if volume leveling is enabled.
We're using 0dB as our guess for video and -10dB for audio.
-
Wow the explanations are coming across to me as more complex than I expected so I must be missing something. Let me try to post the Q again (I was hoping for a short Yes, or No --> set it up like this).
What I want to do is the ability to set the new config in MC once and leave it for ever. Happy to analyse the Audio again and will do so on import.
- If the tracks are from a mixed playlists I want to have their volume levelled across all tracks
- If the tracks are from one album then I don't want volume levelled on a per track basis but on the whole album instead
- I expect it will be the same behaviour for my Music Video as for Music Audio files
- I don't expect I will vol analyse movies (but can if needed) and I don't want MC dynamically adjusting volume during playback of the movie.
I currently have set MC as follows:
- Volume Levelling and Adaptive Volume (Peak Level Normalisation)
-
If you want things to sound the same volume, turn on Volume Leveling.
If you're alright with different playlists being different volumes, turn on Adaptive Volume in Peak Level Normalization mode. It will make the playlist louder, but also cause different playlists to be different volumes.
-
I've Analyse Audio on my 8K Audio files and it finished without any fuss, but some more Q's:
- There are now no values in Replay Gain (I had read a comment that these were still going to be calculated off R128 values for compatibility with other progs)
- Volume Level Album (R128) is blank (for all but one Album)
- Values are not copied to the Stack Members
In some quick listening test I think Volume Levelling and Adaptive Volume (Peak Level Normalisation) gives me what I want. I can see when playing an album the values net out to 0 for each track when they are played and when in a VA smart list each track is adjusted individually.
-
FYI - in testing all of this I solved a long standing annoyance with my "Main PC" that uses a FiiO E10 DAC. I was getting odd "fade outs" in SPL and some clicks and pops. Anyway after mucking around it turns out changing the USB port was all that was needed. Go figure ::)
-
OK - Next dumb Q on this. I have a Various Artist Top 40 "Album" that has been created from different sourced tracks (but it is not a playlist). I want all of these to be use Track not Album Vol Normalisation. As this goes against the rules above, how would you do this? Would I have to use MC's Convert Format function (with DSP --> Vol Leveling enabled) to create a normalised "Album" of this tracks? Thankfully all the Top 40 stuff is probably all mixed to the same hot levels anyway.
-
OK - Next dumb Q on this. I have a Various Artist Top 40 "Album" that has been created from different sourced tracks (but it is not a playlist). I want all of these to be use Track not Album Vol Normalisation. As this goes against the rules above, how would you do this? Would I have to use MC's Convert Format function (with DSP --> Vol Leveling enabled) to create a normalised "Album" of this tracks? Thankfully all the Top 40 stuff is probably all mixed to the same hot levels anyway.
I was wondering the same thing. I have compiled "albums" for several top 40 charts from different years. As I used to use track based leveling this wasn't a real issue but this will probably now be treated as album based?
-
OK - Next dumb Q on this. I have a Various Artist Top 40 "Album" that has been created from different sourced tracks (but it is not a playlist). I want all of these to be use Track not Album Vol Normalisation. As this goes against the rules above, how would you do this? Would I have to use MC's Convert Format function (with DSP --> Vol Leveling enabled) to create a normalised "Album" of this tracks? Thankfully all the Top 40 stuff is probably all mixed to the same hot levels anyway.
I was wondering the same thing. I have compiled "albums" for several top 40 charts from different years. As I used to use track based leveling this wasn't a real issue but this will probably now be treated as album based?
This discussion really belongs in volume leveling rather than adaptive volume, but thinking about this I'm not sure that there is a good automatic solution, and I don't know what the best way would be to set this manually.
If these were simply playlists of files taken from other albums, they would play in track mode automatically because [Album] would be different.
But if you are putting them all in the same folder and calling them an album (e.g. "Top 40, 2010") then they're going to be treated as an album.
You can't algorithmically determine whether this is a "real" compilation album, or if it is a "fake" album that is just a collection of tracks.
If you are checking to see whether [Artist] and [Album] are the same for all tracks and trying to guess that way, you will have too many false positives when looking at real compilation albums, tracks which have multiple artists, or albums which have a guest track on them.
I have compilation albums where the tracks are at specific volumes and crossfade into each other, so using track-based gain would be disastrous.
What were you doing about this in MC18?
Were you just playing all albums in track-mode?
Were you playing to a separate zone?
Were you changing the option manually each time?
-
What were you doing about this in MC18?
Were you just playing all albums in track-mode?
Were you playing to a separate zone?
Were you changing the option manually each time?
Default was track mode, I switched if it bothered me when I was playing an album that was obviously off.
-
I was not addressing this at all in MC18.... but this topic has got me thinking. I presume that the only real way to deal with this is to collect all the tracks together and remaster them as a new album using MC's Convert Format function (with DSP --> Vol Leveling enabled) to create a normalised "Album" of these tracks then delete the "old" collection and use the newly remastered album.
-
I was not addressing this at all in MC18.... but this topic has got me thinking. I presume that the only real way to deal with this is to collect all the tracks together and remaster them as a new album using MC's Convert Format function (with DSP --> Vol Leveling enabled) to create a normalised "Album" of these tracks then delete the "old" collection and use the newly remastered album.
I don't really like that as a solution - normalization should never be touching the audio data. Just think if you had done that for ReplayGain and now wanted to switch to R128.
The only good solutions for this that I can come up with are to either have a special tag or keyword that puts the album into track mode, or to bring back the options, letting the user select "Auto/Track/Album"
I don't like the idea of bringing back the volume leveling options.
Something like this for "Track Gain" seems like it would be best: http://yabb.jriver.com/interact/index.php?topic=80850.0 (http://yabb.jriver.com/interact/index.php?topic=80850.0)
-
So from a use case basis I have:
1) Tracks from with a single "Album" tag where you want only Album based adaptive Vol (eg a rip of Pink Floyd the Wall)
2) Tracks form a playlist consisting of tracks from multiple "Album" tags where you want Track based adaptive Vol
From my testing both of these are currently OK.... but I also have
3) Tracks copied from multiple sources to create a single "Album" tag (eg Hits of the 90's) that also need to have Track based adaptive Vol
Thinking out loud (and to your point) you could have a tag that says "use track based" or else you would have to remaster them using MC's Convert Format function (with DSP --> Vol Levelling enabled) to create a normalised "Album" of these tracks.... There may be a third option I've not thought of....
-
I like this but would rather night mode was a checkbox, with the a couple extra time boxes where you could set start hour, end hour..
-
I like this but would rather night mode was a checkbox, with the a couple extra time boxes where you could set start hour, end hour..
You can create a Night movie zone and put the following in as one of the ZoneSwitch rules (thanks to MrC):
[=Compare(FormatDate(Now(), H), >, 20)]=1
Just change the "20" with the time you want it to start. It is in 24 hour time so 20 equates to 8 pm. If using Internal Volume, you can also set a lower maximum volume for this zone. Move this zone up to a higher priority over the other zones so that if the time is less than your set time, the zone will be ignored and ZoneSwitch will move to the next video zone you have configured.
-
Yeah, seen that thread before still have to try it out for myself.. Still though for an option such as Night Mode, would still make some sense to have native time switching especially for novice users that haven't dabbled in zone switching yet nevermind zone switching with expressions..
You can create a Night movie zone and put the following in as one of the ZoneSwitch rules (thanks to MrC):
[=Compare(FormatDate(Now(), H), >, 20)]=1
Just change the "20" with the time you want it to start. It is in 24 hour time so 20 equates to 8 pm. If using Internal Volume, you can also set a lower maximum volume for this zone. Move this zone up to a higher priority over the other zones so that if the time is less than your set time, the zone will be ignored and ZoneSwitch will move to the next video zone you have configured.
-
Yet another newbie Q: Clip Protection Vs Flat Line. I've got Vol Levelling and Peak Level Normalise on + reanalysed the library. So shouldn't this remove the need for the Clip Protection Vs Flat Line Overflows settings as on some tracks I see an additional reduction when Clip Protection is selected.
Thanks
Nathan
-
How do I disable this feature? ?
It seems to be enabled by default when I play video files.
I am using MC 18.0.211
:-\
-
This feature is new in the upcoming MC19, its impossible to be activated for videos in MC18, you must be seeing something else.
-
How do I disable this feature? ?
It seems to be enabled by default when I play video files.
I am using MC 18.0.211
Adaptive Volume is set in Tools > Options > Video > Adaptive Volume in Media Center 18.
-
Very interesting, a "must-get-19"-feature ;D So I bought 19. ;D :D
-
Peak Level Normalize
If the files being played have not been analyzed using the Analyze Audio tool, the peak level will be learned during playback. In this mode, Audio Path will report "peak level normalize (adaptive)" and the blue light will not illuminate. When used in this mode, the first loud part of a movie or song may cause the volume to be turned down. Once it has turned down, the peak level has effectively been learned and the volume will not be adjusted (unless something even louder happens).
Great example of integrating current technology into the platform.
Before I get started, let me assure everyone that I'm not a blue light moth or a "bit perfect" nut. OK.
Prelude: I rescanned all my audio files with MC19 Audio Analyser and let it run over night and during a work week day, so all my audio has all the new audio analysis tags.. sweet! 8)
Q1. Reading Matt's intro above, I would expect to see the magic blue light on because the audio has been re-analysed but that's not the case. Used to light up in MC18 but "no glow" in MC19 so far.
[For info, DSP is set up as follows-
Output Format (checked) : Output Encoding - None, Channels 2.0 with JRSS
Volume Levelling (checked)
Adaptive Volume Analyser (checked) : Peak Level Normalise (checked)
... that's it, nothing else is on
Audio device is HDMI - WASAPI (default settings) - with this deal in MC18 it was a blue light all the way]
Q2. What's the influence of Volume Mode in all of this?
http://wiki.jriver.com/index.php/Volume#Mode_Details (http://wiki.jriver.com/index.php/Volume#Mode_Details)
I've gone for disabled to avoid internal volume processing and leave it to the external device (in my case the amp) but is it important or just not relevant?
Thanks ..
-
I've tested Adaptive volume with several albums and mixed playlists. I don't like it because it audibly changes the volume, sometimes halfway into a track. For Music listening, Peak Level Normalize seems to be the only viable option as its supposed to preserve dynamic range but on several occasions I could audibly hear it adjust the volume down.
Peter Gabriel's album 4 (security) is a good example. When adding the album in order, you can hear it adjust volume down on the first track because its starting too loud. The song is very quiet and builds up. If you ask me if it has to adjust the volume along the way, it isn't preserving dynamic range.
I probably don't understand how it works, but when you've analyzed an album and you know what the loudest peaks are shouldn't it be easy to determine the maximum gain you can apply? I thought this was how the album mode worked with replaygain in MC18 (and respectfully, I thought that's how peak level normalize should work from the description). If this is true, then I believe peak level normalize is broken.
Fixed! Read my reply below (http://yabb.jriver.com/interact/index.php?topic=82023.msg565313#msg565313).
-
I've tested Adaptive volume with several albums and mixed playlists. I don't like it because it audibly changes the volume, sometimes halfway into a track. For Music listening, Peak Level Normalize seems to be the only viable option as its supposed to preserve dynamic range but on several occasions I could audibly hear it adjust the volume down.
You need to have analyzed your files to avoid this happening. If they have been analyzed, you should not have volume fluctuations (at least not if the playlist is unchanged) - does the Audio Path indicate that Media Center has reduced the volume to avoid clipping?
I do agree that it would be nice if the higher levels of "adaptive volume" were fixed dynamic range compression adjustments rather than dynamic ones though. (not that I'll be using them anyway - it just seems like that would be more consistent?)
-
Everything is analyzed and you're right, its clipping :-[
It just dawned on me why too; the order of processing was incorrect; equalizer came after adaptive volume. When I upgraded to MC19 adaptive volume right was placed right after volume leveling, I never gave it a thought. Both equalizer and room correction came after it. And since it was doing a +4dB fixed, the equalizer and room correction caused it to clip. Disabling adaptive volume removed the +4dB so it wasn't clipping anymore which led me to believe adaptive volume was the problem.
Sorry for the false alarm!
-
Nice to see this feature, as I stream music, and the gain was lowered (-010 db) with this recent change in MC18:
http://yabb.jriver.com/interact/index.php?topic=81245.0
-
Hi Matt,
Fantastic feature I must say, . Does enabling both Adaptive (normalize) and leveling affect music such as classical where passages vary? Im listening to Daft Punk - Giorgio by Moroder, with both features turned on, and i can swear the Modular synth has increased in volume, or is that just my imagination?
Aside, just an idea..
Whilst re-analysing my library to fit in with the new Adaptive volume feature, which is quite huge btw (8000 tracks), the log is actively scrolling and there seems to be no way to stop it other than fighting with it - yes, I always loose. I've got JRiver analysing atm whilst listening to music, it's taking a fair while even with 4 processes enabled. I would like to keep track of what has been analysed whilst I'm listening to music so I can play only those that have finished analysing.
How about adding a pause scroll log 'checkbox', so the user can scroll freely and see what's already been done.
-
Hello Matt,
I've tested Adaptive volume with several albums and mixed playlists. I don't like it because it audibly changes the volume, sometimes halfway into a track. For Music listening, Peak Level Normalize seems to be the only viable option as its supposed to preserve dynamic range but on several occasions I could audibly hear it adjust the volume down.
Same behavior here. All files are analyzed. Volume changes while tracks are playing. Look at this (https://dl.dropboxusercontent.com/u/80739830/bandicam%202013-08-24%2015-41-50-976.avi) captured File. Start/Stop/Start Files with JRemote, and captured the behavior in JRiver. Please look for the the adaptive volume changing on screean while playing...
Thanks!
-
Apart from the fact that I also made a mistake with DSP order, there's also been a bug confirmed (and fixed in the next release) (http://yabb.jriver.com/interact/index.php?topic=83010.0).
-
Hello Matt,
Same behavior here. All files are analyzed. Volume changes while tracks are playing. Look at this (https://dl.dropboxusercontent.com/u/80739830/bandicam%202013-08-24%2015-41-50-976.avi) captured File. Start/Stop/Start Files with JRemote, and captured the behavior in JRiver. Please look for the the adaptive volume changing on screean while playing...
It looks like these files were analyzed by a previous version of Media Center - you need to re-analyze files in MC19 for volume leveling and adaptive volume to work correctly. Older versions of Media Center did not perform true peak level analysis. (Peak Level R128)
-
It looks like these files were analyzed by a previous version of Media Center - you need to re-analyze files in MC19 for volume leveling and adaptive volume to work correctly. Older versions of Media Center did not perform true peak level analysis. (Peak Level R128)
The Files were analyzed with MC 19.0.27. should i reanalyse them again?
-
Can you check whether your files have peak level (R128) and volume level (R128) tags set? If they are, you could try reanalysing 1 or 2 files that fluctuate and see if that helps.
-
The audio path is showing a mix of "adaptive" and "fixed" files even though I've analyzed the entire library. What could cause some files to show peak level normalize(fixed) and others peak level normalize(adaptive)? I'm confused, any help is appreciated.
-
The audio path is showing a mix of "adaptive" and "fixed" files even though I've analyzed the entire library. What could cause some files to show peak level normalize(fixed) and others peak level normalize(adaptive)? I'm confused, any help is appreciated.
There's a bug when the files have a peak of 0.0. It will be fixed next build.
-
I've tested Adaptive volume with several albums and mixed playlists. I don't like it because it audibly changes the volume, sometimes halfway into a track.
I've noticed this as well... it reminds me of the days of recording onto a cassette tape from a tape deck that had auto-levels instead of setting whatever level you wanted with a knob (ah, the old days! 8) ). Music would start off too loud and then fade down.
What I've discovered that works decent for me is to add a filter in the Parametric EQ to add +10db to the output while using volume leveling. This seems to compensate for tracks that were mastered in the -14db range all the way to the -4db range without a lot of clipping/flat line overflow. I'll continue experimenting. :) It's why I upgraded, just to experiment! ;D
-
I've noticed this as well... it reminds me of the days of recording onto a cassette tape from a tape deck that had auto-levels instead of setting whatever level you wanted with a knob (ah, the old days! 8) ). Music would start off too loud and then fade down.
What I've discovered that works decent for me is to add a filter in the Parametric EQ to add +10db to the output while using volume leveling. This seems to compensate for tracks that were mastered in the -14db range all the way to the -4db range without a lot of clipping/flat line overflow. I'll continue experimenting. :) It's why I upgraded, just to experiment! ;D
Check the order of DSP items. Where is Adaptive Volume? In my case, Equalizer was below ADaptive volume and caused clipping and since I had clip protection enabled, it audibly turned down the volume.
You can rearrange the items by dragging them; after I dragged Equalizer above Adaptive volume the problem was solved.
-
This feature alone sells me on the upgrade. I've spent a lot of time tweaking a third party VST dynamics plugin to try and get this kind of functionality and have never been all that impressed. How I wished for a MC native version. Now there is. Great work guys!
-
The audio path is showing a mix of "adaptive" and "fixed" files even though I've analyzed the entire library. What could cause some files to show peak level normalize(fixed) and others peak level normalize(adaptive)? I'm confused, any help is appreciated.
Do all files have a Peak Level (R128) value?
-
The majority do. I have ~7,500 files and I show 9 that didn't get analyzed due to "error". I'm still working on the nature of the error since the files appear fine and play normal.
I ran a smart list of files with 0.0db after I saw Matt's response. That list was only 62 files. I have many more files than the 62 that show adaptive mode in the audio path, so I'm not sure if the bug fix that Matt referenced addresses my particular issue.
Thanks for any suggestions.
-
Check the order of DSP items. Where is Adaptive Volume?
Output Format
Volume Leveling
Adaptive Volume
Parametric EQ (when active)
Still has that effect of blast the quiet parts and hurry-up-and-quiet-down the loud parts.
-
I use Adaptive Volume in "Peak Level Normalize" mode, and i never noticed the volume changing mid-playback.
You aren't using "Night Mode", are you? Because that is designed to boost quiet parts and quiet down loud parts, specifically.
-
Output Format
Volume Leveling
Adaptive Volume
Parametric EQ (when active)
Still has that effect of blast the quiet parts and hurry-up-and-quiet-down the loud parts.
What is PEQ doing? In my case, the regular equalizer was boosting by +6dB on preamp slider (leftover from the past where everything was too quiet). Adaptive volume takes care of that now so that was causing it to clip. But boosting individual freqs also boost the total of the signal, so that could also cause a clip.
Turns out though I have to eat my own words about the place of the EQ. I tested this again and the order didn't matter for the regular EQ, boosting +6dB while above Adaptive volume also caused a clip. In my defense, this is not what the footnote says in the DSP dialog (ie, processed in order). So I wonder, is that by design or is it a bug?
-
Could be a design limitation, possibly a bug.
Adaptive Volume uses the Peak tags from the audio analysis, however if you boost the audio, those peak values are wrong. Not sure Adaptive Volume can really know that the audio was boosted before, too many factors can influence that, really. Don't boost the audio if you want to use adaptive volume? =)
-
Don't boost the audio if you want to use adaptive volume? =)
Yes, of course you're right. Before adaptive volume was introduced (and pre-R128) using RG only, my output using ASIO was always ~6dB lower than with Wasapi so I had a fixed +6dB configured in the old DSP dialog where you would choose track or album mode. When this was changed, RG values were removed and that dialog changed but I didn't immediately reanalyzed everything for use with R128, so I resorted to the preamp slider in the EQ until I was done reanalyzing everything. I tend to forget things like that so it started clipping once everything was done :).
-
What is PEQ doing? In my case, the regular equalizer was boosting by +6dB on preamp slider (leftover from the past where everything was too quiet). Adaptive volume takes care of that now so that was causing it to clip. But boosting individual freqs also boost the total of the signal, so that could also cause a clip.
Turns out though I have to eat my own words about the place of the EQ. I tested this again and the order didn't matter for the regular EQ, boosting +6dB while above Adaptive volume also caused a clip. In my defense, this is not what the footnote says in the DSP dialog (ie, processed in order). So I wonder, is that by design or is it a bug?
I have tried it with the PEQ unchecked. It seems to work great for albums in this configuration, but in playlists is when it's most noticeable. And I'm using Peak Level Normalize.
-
Keep audio path open while you're playing that playlist (hover your mouse over the DSP button for the popup). This will show exactly what is happening. Under Changes, you should see a number fluctuate if MC is responsible for the volume fluctuations. In my case, it showed 'Adjust volume for clip protection'.
-
I'm at work right now but I will test this out tonight when I get home. I have seen it fluctuate however, especially on a track like Yes' "Close To The Edge" where it starts off really quiet and builds up at the opening. I can't remember what the values are but I'll try again tonight and let you know.
-
Try making a screenshot of it and mention which line it is you see fluctuating.
Also, lets not forget that
There's a bug when the files have a peak of 0.0. It will be fixed next build.
-
I may have found the bug that was causing this:
- Try enabling Adaptive Volume and Volume Leveling at the same time, then play a track.
- Stop playback, and disable Volume Leveling.
- Adaptive Volume should now work correctly for analyzed files.
-
Try making a screenshot of it and mention which line it is you see fluctuating.
(http://www.mgdking.com/images/audiopath.png)
The second line fluctuates. Also, I've noticed that some tracks say fixed while others say adaptive. And they can be on the same album.
(http://www.mgdking.com/images/dspstudio.png)
This is my audio chain.
I noticed that the sound didn't fluctuate as bad as the night before. Maybe that's due to this:
I may have found the bug that was causing this:
- Try enabling Adaptive Volume and Volume Leveling at the same time, then play a track.
- Stop playback, and disable Volume Leveling.
- Adaptive Volume should now work correctly for analyzed files.
-
The second line fluctuates. Also, I've noticed that some tracks say fixed while others say adaptive. And they can be on the same album.
Did you check whether those tracks had a peak of 0,0? That could be the bug Matt spoke about.
If not then maybe its something to do with what 62 found out, I don't know.
-
(http://www.mgdking.com/images/audiopath.png)
The second line fluctuates.
Have a look at my post #45, there is a video capture with this behavior
https://dl.dropboxusercontent.com/u/80739830/bandicam%202013-08-24%2015-41-50-976.avi (https://dl.dropboxusercontent.com/u/80739830/bandicam%202013-08-24%2015-41-50-976.avi)
All files are analyzed with the current version of MC 19. the Problem is still there with playlists without peak 0.0 files
-
All files are analyzed with the current version of MC 19. the Problem is still there with playlists without peak 0.0 files
I believe the fix is in build 19.0.33, which should be public next week.
Thanks.
-
Great, thank you Matt!
-
Can the initial Analyzing Audio process be run on one machine and restored via Library restore to another? I want to run it on my i5 main system. The AMD HTPC machine is sloooww at analyzing audio.
-
You are mentioning a Blue light, I just have re-analyzed all my library 40K songs, and have set the volume leveling and the Adaptive Volume to work together, Peak Level Normalize is (Fixed) but cannot see any Blue Light, What do you mean with the blue light, how can I enable it?
Thanks
Jherm
Overview
Adaptive Volume is a system to keep the level of the input signal near full level.
There are several use-cases where this feature is useful:
- You use Volume Leveling and don't want the volume of everything turned down a lot unless it's actually necessary to level the volume of the current playlist
- You would prefer to run your DAC close to full level on any input (more important for DSD and classical where the signal may never approach full scale)
- You watch movies at night or in a situation where explosions are too loud and dialog is too quiet
- You're trying to get enough volume out of a low power system, like during playback on laptop speakers.
MC18 added an adaptive volume mode available for video playback. MC19 promotes this system to be available for any media type, and improves it in several important ways.
This screenshot shows Adaptive Volume in MC19 and explains the modes a little bit:
(http://files.jriver.com/images/2013/adaptive_volume.png)
Peak Level Normalize
This mode requires special consideration. It analyzes the current playlist to find the peak level of the entire playlist by using the values gathered during Audio Analysis. It also considers the peak level of the entire playlist if Volume Leveling is engaged. Then, it normalizes the playlist to this peak level. Since this applies a fixed gain, the blue light will illuminate and Audio Path will say "peak level normalize (fixed)".
If the files being played have not been analyzed using the Analyze Audio tool, the peak level will be learned during playback. In this mode, Audio Path will report "peak level normalize (adaptive)" and the blue light will not illuminate. When used in this mode, the first loud part of a movie or song may cause the volume to be turned down. Once it has turned down, the peak level has effectively been learned and the volume will not be adjusted (unless something even louder happens).
Let's describe a simple example of why this mode might be useful. If you play Adele's 21 album, a modern best-seller mastered in a modern way, and enable Volume Leveling, it will request that all songs be turned down by 14.4dB. This is so that if you play something mastered more conservatively, it will match Adele's volume. However, if you are _only_ playing Adele, you're going to be applying the same -14.4dB to all the songs you hear. You didn't actually need to turn the music down at all to achieve a level volume. Peak Level Normalization will understand this and add the 14.4dB back (internally it's smart enough to do nothing instead of doing two offsetting changes).
Said another way, if you use Volume Leveling and Peak Level Normalization, you will get the loudest playback of the current playlist that maintains equal volume between tracks and prevents all clipping. The volume between tracks in the playlist will be the same, but the volume between different playlists could be different (since each playlist will have a different peak level normalization value).
-
You are mentioning a Blue light, I just have re-analyzed all my library 40K songs, and have set the volume leveling and the Adaptive Volume to work together, Peak Level Normalize is (Fixed) but cannot see any Blue Light, What do you mean with the blue light, how can I enable it?
You need to be using the default Noire theme, or a theme which supports it:
(http://www.abload.de/img/bitperfectbwup7.gif)
-
You need to be using the default Noire theme, or a theme which supports it:
(http://www.abload.de/img/bitperfectbwup7.gif)
Thanks
-
I'm having the same issue with audio analysis not changing the behavior of adaptive volume (I'm using peak normalization mode). I successfully analyzed a video (in MC19), and I enabled the R128 columns in the library to confirm that the file had been tagged with the audio analysis data. It does not have a peak value of 0.0 (the closest it got was -3.0 on the center channel), but the adaptive volume behavior is exactly the same as it was when playing without having run the audio analysis first. The volume boost starts out very high (like +12db) but fairly quickly drops down as new volume peaks are detected, causing noticeable shifts in the volume level. There is only this one file in my library and only one file on the playlist. Volume leveling is not enabled.
Looking forward to the .33 build.
-
I'm having the same issue with audio analysis not changing the behavior of adaptive volume (I'm using peak normalization mode). I successfully analyzed a video (in MC19), and I enabled the R128 columns in the library to confirm that the file had been tagged with the audio analysis data. It does not have a peak value of 0.0 (the closest it got was -3.0 on the center channel), but the adaptive volume behavior is exactly the same as it was when playing without having run the audio analysis first. The volume boost starts out very high (like +12db) but fairly quickly drops down as new volume peaks are detected, causing noticeable shifts in the volume level. There is only this one file in my library and only one file on the playlist. Volume leveling is not enabled.
Looking forward to the .33 build.
Then I removed the file from the library, re-imported it, re-ran the audio analysis and now it works as expected. A fixed +2.0db gain from the very beginning...
-
Adaptive volume also seems to be effected by other files in the playlist:
- File 1 has a peak volume R128 level of -3.0. When played by itself, adaptive volume adds a fixed boost of +2.0db.
- File 2 has a peak volume R128 level of -2.0. When played by itself, adaptive volume adds a fixed boost of +1.0db.
When both files are on the Playing Now list, playing File 1 now has a fixed boost of +1.0db, but when it was played by itself it had a boost of +2.0db. I thought balancing volume levels across a playlist was the responsibility of the Volume Leveling feature? I do not have the Volume Leveling feature enabled.
EDIT: I re-read Matt's original post and now I see this is intended behavior. My question is, why? In my opinion this should only be the behavior if Volume Leveling is enabled. If Volume Leveling is not enabled, then Adaptive Volume should boost on a per-file basis regardless of what other files are in the playlist. Adjusting on a per-playlist and not a per-track basis seems to totally defeat the point of this plugin...imagine having two files on the playlist, one of which is very loud and the other is very quiet. I'd expect a feature called "adaptive volume" to make the quiet file much louder. Which it would, when played by itself. But due to current design which considers the peak volume at the playlist level, when played in such a playlist the quiet file will receive hardly any boost at all. This seems completely counter-intuitive to the description of "Peak level normalize" given in the plugin which makes no references at all to peak playlist levels.
EDIT 2: I see Matt's response to this in a previous post:
He could always play one track at a time ;)
Seems a rather poor solution... A much better solution, like I mentioned before, would be to make adaptive volume work on a per-file basis when used alone and only consider peak playlist level when enabled in conjunction with volume leveling.
EDIT 3: I think I understand now why this was designed the way it was. It was designed with only music album playlists in mind. So the current behavior does make some sense because it preserves the (often artistically intentional) dynamics in transition from one track to the next. But this ONLY applies to music albums. For other types of playlists such as mixed music playlists or video playlists there is no intentional transition to preserve so the way the plugin is designed makes much less (none, really) sense.
-
That behavior seems like a bug. I would have to go back and check, but I'm guessing it may have been introduced back when the way that album leveling works was changed. What should be happening:
Adaptive Volume enabled
All tracks are individually adjusted to play back at the maximum level, whether they are part of an album or not. Loudness varies for each track.
Adaptive Volume and Volume Leveling enabled
Volume Leveling is applied, and the playlist is adjusted to play at the maximum level possible. (determined by the track with the highest peak level in the playlist)
Loudness should be equal across each track, but different playlists may have different loudness levels.
Volume Leveling Enabled
Individual tracks are normalized to -23 LUFS.
Albums should have their average level normalized to -23 LUFS. (albums currently use the peak volume, which means they usually play back too quietly - this was a serious regression in build 14)
As long as this provides sufficient headroom (it should in most cases with music) loudness should be equal across all tracks and playlists - you should never have to change the volume control.
Special Cases
Videos always use track-based volume leveling, unless their Media Sub Type is Music Video.
Audio which has a Media Sub Type of Podcast or Radio should always use track-based volume leveling.
There needs to be some way to mark an album as a "Mixtape" to force track-based leveling.
There needs to be some way of reducing the volume level target when -23 LUFS does not provide enough headroom. (typically videos, rather than music) Personally I would like to see this linked to the internal volume control.
-
That behavior seems like a bug. I would have to go back and check, but I'm guessing it may have been introduced back when the way that album leveling works was changed. What should be happening:
Adaptive Volume enabled
All tracks are individually adjusted to play back at the maximum level, whether they are part of an album or not. Loudness varies for each track.
Adaptive Volume and Volume Leveling enabled
Volume Leveling is applied, and the playlist is adjusted to play at the maximum level possible. (determined by the track with the highest peak level in the playlist)
Loudness should be equal across each track, but different playlists may have different loudness levels.
Volume Leveling Enabled
Individual tracks are normalized to -23 LUFS.
Albums should have their average level normalized to -23 LUFS. (albums currently use the peak volume, which means they usually play back too quietly - this was a serious regression in build 14)
As long as this provides sufficient headroom (it should in most cases with music) loudness should be equal across all tracks and playlists - you should never have to change the volume control.
I agree 100%. That is exactly how I would expect them to function. You explained it very well.
-
That behavior seems like a bug. I would have to go back and check, but I'm guessing it may have been introduced back when the way that album leveling works was changed. What should be happening:
Adaptive Volume enabled
All tracks are individually adjusted to play back at the maximum level, whether they are part of an album or not. Loudness varies for each track.
Adaptive Volume and Volume Leveling enabled
Volume Leveling is applied, and the playlist is adjusted to play at the maximum level possible. (determined by the track with the highest peak level in the playlist)
Loudness should be equal across each track, but different playlists may have different loudness levels.
Volume Leveling Enabled
Individual tracks are normalized to -23 LUFS.
Albums should have their average level normalized to -23 LUFS. (albums currently use the peak volume, which means they usually play back too quietly - this was a serious regression in build 14)
As long as this provides sufficient headroom (it should in most cases with music) loudness should be equal across all tracks and playlists - you should never have to change the volume control.
To me, using Adaptive Volume (I will always use Volume Leveling) has no value, since it would mean the volume a particular track plays depends on what else is in the current playlist. So for example, I'm playing PlayingNow... and add more tracks, the volume could change... My goal is to never have to touch the volume control as albums and tracks change and as I maintain playlists.
-
To me, using Adaptive Volume (I will always use Volume Leveling) has no value, since it would mean the volume a particular track plays depends on what else is in the current playlist. So for example, I'm playing PlayingNow... and add more tracks, the volume could change... My goal is to never have to touch the volume control as albums and tracks change and as I maintain playlists.
I don't use it either. Some people complain that volume leveling is too quiet, so using the two gives you level playback for the current playlist, and makes it as loud as possible.
I'd rather just turn up the volume on my amp and keep it level all the time.
-
In a coming build:
Changed: Adaptive Volume in 'Peak Level Normalize' mode normalizes on a per-track basis when Volume Leveling is off and on a per-playlist basis when Volume Leveling is on.
-
In a coming build:
Changed: Adaptive Volume in 'Peak Level Normalize' mode normalizes on a per-track basis when Volume Leveling is off and on a per-playlist basis when Volume Leveling is on.
Great news Matt. I was waiting off on the MC19 upgrade to see what came of this and now I have no excuse :)
-
Seeing as running Volume levelling + Adaptive Volume together pushes the sound as far as it will go, without clipping, would it make sense to Disable the Volume and use my amp to control the final output? Are there any downsides to lowering the internal with both features enabled?
(https://dl.dropboxusercontent.com/u/20406996/jriver%20path.png)
-
Seeing as running Volume levelling + Adaptive Volume together pushes the sound as far as it will go, without clipping, would it make sense to Disable the Volume and use my amp to control the final output? Are there any downsides to lowering the internal with both features enabled?
There are no problems with using the internal control, but I don't see the point in setting up Media Center to play as loud as possible, and then reducing the volume.
Are you sure you don't just want Volume Leveling enabled on its own?
-
Volume Leveling Enabled
Individual tracks are normalized to -23 LUFS
Would it make sense, while retaining the -23LU target as a default, to permit the actual target loudness to be user defined?
-
Volume Leveling Enabled
Individual tracks are normalized to -23 LUFS
Would it make sense, while retaining the -23LU target as a default, to permit the actual target loudness to be user defined?
I do think that the target volume should change if Media Center's internal volume control is adjusted. (so that -7 dB would be -30 LUFS for example) This way when you reduce the volume control, you gain additional headroom for volume leveling - which would be useful for video playback.
Using a louder target than -23 LUFS does not provide enough headroom for all tracks to be sufficiently normalized, and is not compliant with the R128 spec.
That's why the recommendation is to enable both adaptive volume (set to peak level normalization) and volume leveling at the same time, if you want to make use of volume leveling but your amplifier can't go loud enough with it enabled.
Having the two enabled like this levels the current playlist, and then increases the volume as much as possible without clipping. Depending on what's playing, this may result in a small increase in volume, or a larger increase in volume compared to using volume leveling on its own. Volume across the current playlist will be level, but the trade-off is that volume will vary as the playlist is changed.
-
Hey, is there a way that Adaptive volume can be enabled automatically for video playback only?
If not is it possible for this feature to be added please?
-
Hey, is there a way that Adaptive volume can be enabled automatically for video playback only?
If not is it possible for this feature to be added please?
You'll have to create two zones, and then use the Zone Switch feature to switch between the zones for audio and video - that way you can have separate configuration.
-
Thanks, although from memory this would not be required (creating 2 zones) with the equivalent/similar Adaptive Volume functionality in version 18.
In version 18 Adaptive Volume was only functional when playing video files.
-
I've noticed adaptive volume with volume leveling handles "Play Doctor" playlists differently than user created playlists. Is this normal? What I've seen is a play doctor playlist created in a blank pane on start up would use a per track peak level evaluation(adaptive) rather than per playlist evaluation(fixed). If you close MC and re-open the user interface then start playing a selection from the play doctor playlist, MC will use a per playlist peak level evaluation(fixed). I'm using build 38....other playlist conform to the per playlist evaluation as expected.
Anyone else notice this?
-
I've noticed adaptive volume with volume leveling handles "Play Doctor" playlists differently than user created playlists. Is this normal? What I've seen is a play doctor playlist created in a blank pane on start up would use a per track peak level evaluation(adaptive) rather than per playlist evaluation(fixed). If you close MC and re-open the user interface then start playing a selection from the play doctor playlist, MC will use a per playlist peak level evaluation(fixed). I'm using build 38....other playlist conform to the per playlist evaluation as expected.
Anyone else notice this?
I would not recommend enabling both these options when using Play Doctor, because its playlist is dynamic. Use one or the other.
-
I would not recommend enabling both these options when using Play Doctor, because its playlist is dynamic. Use one or the other.
Thanks.
-
A feature I'd like to see added is allowing the user to specify the max boost. This would be especially helpful for the Night Mode setting.
Sometimes, the +20db max boost is too much. Background noise can get very loud and noticeable.
Perhaps make +20 the highest allowed, but it would be useful to allow the user to specify any max boost between 1-20db.
-
1. It would be better if adaptive volume goal would be -0.2db and not 0db (a slightly less than maximum possible gain addition without clipping).As all audio is now dithered to the output bitdepth format , and dither noise can add a slight boost to gain , it is possible to end up with clipping audio after all.
2. In DSP chain appear two lines : one for volume leveling and the other for adaptive volume. Are these calculated in two stages or just appear this way for readability? Wouldn't it be better for processing audio in minimum possible stages , ie to join the 2 stages into one processing after calculating the result?
3. When converting audio format , is adaptive volume calculated correctly as in regular playback?
-
1. It would be better if adaptive volume goal would be -0.2db and not 0db (a slightly less than maximum possible gain addition without clipping).As all audio is now dithered to the output bitdepth format , and dither noise can add a slight boost to gain , it is possible to end up with clipping audio after all.
Volume Leveling in MC uses -1.0dB True Peak as a maximum.
Even with Volume Leveling disabled, dither won't cause a clip in JRiver.
2. In DSP chain appear two lines : one for volume leveling and the other for adaptive volume. Are these calculated in two stages or just appear this way for readability? Wouldn't it be better for processing audio in minimum possible stages , ie to join the 2 stages into one processing after calculating the result?
When possible, all fixed gains are applied in one stage (using SSE acceleration).
With regards to audio quality, it doesn't matter if you do lots of little changes or one big change due to JRiver's 64-bit precision. Proof here (http://wiki.jriver.com/index.php/Audio_Bitdepth#Bit-Perfect).
-
As volume leveling is available for in DLNA server configuration, can we expect to get adaptive volume available in this DLNA server configuration?
For my part I'm using DLNA server of JRiver MC 19 to play music on my Pure Evoke F4 radio but when I activate adaptive volume music sound too low on this radio.
Volume is OK when I is not activated. Only non compressed/expanded tracks are sounding too low.
I think that the possibility to boost the sound using DLNA server of JRiver could be useful.
-
Is there anyway you could get the nightmode to come on and go off automatically at set times?? Or setup a keyboard shortcut?
-
Is it possible to include the option of disabling the album feature completely, as it was on previous versions? My album tag has nothing to do with a real album.
What I want Is volume levelling according to analysis and boost with adaptive volume, but without any influence of the album name.
-
Is there anyway you could get the nightmode to come on and go off automatically at set times?? Or setup a keyboard shortcut?
You might be able to use time-based zone switching. I've described it here:
http://yabb.jriver.com/interact/index.php?topic=76605.msg519030#msg519030 (http://yabb.jriver.com/interact/index.php?topic=76605.msg519030#msg519030)
-
It looks like these files were analyzed by a previous version of Media Center - you need to re-analyze files in MC19 for volume leveling and adaptive volume to work correctly. Older versions of Media Center did not perform true peak level analysis. (Peak Level R128)
So, what I'm hearing is that the volume leveling is improved in MC19?
One of my favorite pet peeves with MC was that volume leveling never really worked for me. Some albums low, some blasting, all analyzed. Was hoping that at some time you'd allow manual editing of the replay gain and other options in the ANALYZE AUDIO table. Right now, only options seem to be what the automatic process sets, or "0" (and not even that on some settings). Seems as if that would be a simple switch in whatever database software you're using in the back end. That would allow me to reset an album or track to match the average levels for my tracks that apparently the automatic process can't deal with.
I've been using MC since v16, so would re-analyzing the library maybe make any difference? If not, would upgrading to MC19, AND enabling both volume leveling and adaptive volume leveling AND re-analyzing the library fix these inconsistencies?
(Still be nice to be able to have a manual option when all else fails)
-
So, what I'm hearing is that the volume leveling is improved in MC19?
Volume Leveling has been considerably improved in MC19 - though you may still have audible differences between albums right now.
Currently it uses the gain from the loudest track on an album rather than the average, which can still result in uneven playback.
This is not a problem when you are playing mixed/random playlists though - those take full advantage of the new leveling system.
I've been using MC since v16, so would re-analyzing the library maybe make any difference?
You need to re-analyze your files to take use the new volume leveling system, otherwise it still uses the old ReplayGain v1 method rather than the new R128 method.
If not, would upgrading to MC19, AND enabling both volume leveling and adaptive volume leveling AND re-analyzing the library fix these inconsistencies?
Adaptive Volume creates volume inconsistencies.
If you have the two enabled, playback in the current playlist ("Now Playing" view) is leveled, and then adaptive volume makes that list of tracks as loud as possible without clipping.
This means that your current playlist should have an even volume, but different playlists may play back at different volume levels.
If you enable Volume Leveling on its own, the volume should be level regardless of what you are playing. (as level as is possible, anyway) Everything is normalized to a target of -23 LUFS.
Hopefully the JRiver team will get around to fixing the album leveling soon.
It was fine until 19.0.14, if I recall correctly, and then the change was made from using the average gain to the loudest track to fix a problem with video playback when volume leveling was enabled.
The way that volume leveling applies to videos has changed since then, so that fix should no longer be necessary, but it's still in place and is holding back the performance of the volume leveling.
Even so, it should still be a lot better than the old ReplayGain method.
-
I don't do playlists as a rule - I do albums, and if a track suggests something else, manually pick 'n click to fire that up ... repeat as necessary.
Way I read it, adaptive leveling isn't what I'm looking for, but volume leveling <may> improve the situation here due to the improved processing. Ultimate goal would be to be able to just let the volume control gather dust ...
OK then ... let me munch on that for a bit ...
PS ... the latest STABLE version contain all this goodness?
-
Way I read it, adaptive leveling isn't what I'm looking for, but volume leveling <may> improve the situation here due to the improved processing. Ultimate goal would be to be able to just let the volume control gather dust ...
Adaptive Volume is generally for people that want the loudest playback they can get.
In MC19 it can be combined with Volume Leveling to give you the loudest volume for the current playlist, while leveling it, but if your goal is to never touch the volume control, you should leave Adaptive Volume disabled.
If you just want an even volume and to never have to touch the volume control, that's what Volume Leveling on its own will do.
You may still have to adjust your volume control a little until we get a fix for the way that albums are leveled though.
It seems like a really small change, so I'm not sure why it hasn't been fixed yet.
PS ... the latest STABLE version contain all this goodness?
Yep.
-
Yeah ... seems like you could set a universal reference level for playback and have the volume leveling do what it has to do to match that ... maybe a test track or two. Set those to match your preferences and tell MC ... that's it! NOW do your thing ...
-
Yeah ... seems like you could set a universal reference level for playback and have the volume leveling do what it has to do to match that ... maybe a test track or two. Set those to match your preferences and tell MC ... that's it! NOW do your thing ...
Volume Leveling does this with a target of -23 LUFS (dB) as the R128 specification requires.
I do think it would be beneficial for the target to shift as you reduce volume though, to provide additional headroom for volume leveling with particularly dynamic tracks.
99% of my music library fits with this 23dB range though - it's video playback where leveling can run out of headroom. (which means that some videos are not completely level)
-
Point of order ... should adaptive volume be turned on PRIOR to re-analysis?
I did the upgrade and re-analyze thing last night. Around three hours for 5000 flacs, which seems reasonable. First impression was that everything was much quieter. Huh?
Second impression ... TURN IT DOWN before clicking the adaptive volume on. Crap! Lucky I didn't throw cones when it kicked in. Once I got past that (and cleaned the stains off the chair) the volume going album to album is a lot closer than it was. I went through several selections that I know would have required tweaking in the past, with good results and no reaching for the controls. As with anything new, I expect it will take a bit of getting used to, but it feels like progress.
I suppose one thing I'd add to my wish list is the ability to change that -23db target to match my system and library. That would avoid the "what happens if I click this OMG WE'RE ALL GONNA DIE! issue when turning on adaptive volume ... NO option should have the potential for damage that one currently has, at least here. There too, a test track or two selected and based on average sampling and results might be handy.
-
Point of order ... should adaptive volume be turned on PRIOR to re-analysis?
It won't affect analysis at all, it's only applied during playback.
First impression was that everything was much quieter. Huh?
R128 has a lower target volume than ReplayGain - it's required in order to give it enough headroom to adequately level highly dynamic tracks.
I suppose one thing I'd add to my wish list is the ability to change that -23db target to match my system and library. That would avoid the "what happens if I click this OMG WE'RE ALL GONNA DIE! issue when turning on adaptive volume
If your goal is to never touch the volume control, do not use adaptive volume.
The whole point of adaptive volume is to play at the maximum loudness possible without clipping.
-23dB was chosen for the R128 specification (it is an international standard - not something JRiver has developed themselves) specifically because it allows just about any music to be adequately leveled. If you increase the target loudness from that, you are very likely to run into problems where it starts to sound uneven again.
The only change I would like to see regarding the target level, is to have it change with the volume control.
So -6dB on the volume control would actually change the target to -29dB, giving you an extra 6dB of headroom.
Leveling is currently applied before the volume control, so if you run out of headroom, there's nothing that can be done about it.
-
Did some goooogling, and R128 is a video standard - IMHO ... R128 is WAY too conservative for audio playback for pure listening.
Just played a couple of my favorite "wall of sound" tracks. Wedding Nails by Porcupine Tree and the grand finale of Kapp's Symphony #2. With volume leveling enabled and adaptive volume disabled, I was only seeing peak levels in at around 20% in the equalizer panel ... that's kinda sorta ridiculous, right? What kind of dynamic range we going for here? Even with the preamp level of the equalizer jacked all the way up to 12db, I was seeing 75% peaks at most, which seems a sad waste of potential. I'd prefer to see the clip protection take care of the once in a blue moon excursions that exceed a more agressive setting. Lucky me, I'm tube here, so even without that, I'd be seeing soft clips at worst. Once again, the "cure" might be to just have a master override setting for the -23db base point that R128 uses. Also be good if you could apply a different master compensation setting to video and dedicated audio tracks.
Maybe related - I'm also hearing some audible pumping that wasn't there before. Not sure what to make of that. Possibly I'd need to turn down my trusty old dbx 3BX - stands to reason feeding that a wider dynamic range would require compensation. The visual display doesn't look any different than what I'm used to seeing though.
One thing I'm already missing is the compensation settings in volume leveling that allowed you to boost or cut the overall results for albums or tracks. That could do the trick. See my comment above about a master override setting. Also be nice to be able to select which type of leveling you want to use - the original with all it's baggage, or this one with a new set of baggage.
So maybe it's just me, but it looks like at this point, if I were to decide the old system worked better for me, my only option to go back would be to downgrade back to version 18? Probably sooner the better so I don't stumble across any killer new features in v19 I just can't live without. <G>
-
Did some goooogling, and R128 is a video standard - IMHO ... R128 is WAY too conservative for audio playback for pure listening.
R128 is a broadcast standard (https://tech.ebu.ch/loudness), not a video standard. Radio stations in Europe are already implementing it.
Video frequently requires more than 23dB of headroom in my experience - I'd like to increase it to something around 32dB.
Just played a couple of my favorite "wall of sound" tracks. Wedding Nails by Porcupine Tree and the grand finale of Kapp's Symphony #2. With volume leveling enabled and adaptive volume disabled, I was only seeing peak levels in at around 20% in the equalizer panel ... that's kinda sorta ridiculous, right?
This depends entirely on the dynamic range of your audio. Highly dynamic tracks may even exceed the amount of headroom that R128 provides, but that's about 1% of my library.
If you add an expression column and use this code, it will tell you how much headroom is required for your tracks:
Delimit(if(isempty([Peak Level (R128)]),,formatnumber(math(removecharacters(left([Peak Level (R128)],5),/ /+,0)+RemoveCharacters([Volume Level (R128)],/ LU,0)+1),1)),/ dB,)
As you approach 0 dB you are reaching the limits of what 23dB headroom allows, and you may even find tracks which exceed it.
It may also be useful to add the Dynamic Range (DR) column as well, as that gives you a rough quality analysis of your library. (higher is better, and well mastered tracks are typically above 10)
I'd prefer to see the clip protection take care of the once in a blue moon excursions that exceed a more agressive setting.
R128 should never clip - if it runs out of headroom you simply have a track which does not level correctly. (plays back quieter than the rest)
Maybe related - I'm also hearing some audible pumping that wasn't there before.
Adaptive Volume will do this if it's in anything but the Peak Level mode, or if it's being used with tracks which have not been analyzed.
One thing I'm already missing is the compensation settings in volume leveling that allowed you to boost or cut the overall results for albums or tracks. That could do the trick. Also be nice to be able to select which type of leveling you want to use - the original with all it's baggage, or this one with a new set of baggage.
Volume Leveling should not require these controls.
-
Appreciate your patience here ...
So, with volume leveling on and adaptive off, with the equalizer set for +6db and otherwise flat, and checking the same files I referenced earlier ...
(the envelope please)
Wedding Nails (CD redbook rip to FLAC)
- Dynamic Range (R128)= 16.0 LU
- Dynamic Range (DR) = 7
- Your expression = -11.5dB
Kapp's Second ... (CD redbook rip to FLAC)
- Dynamic Range (R128)= 13.6 LU
- Dynamic Range (DR) = 10
- Your expression = -9.4 dB
I checked out a higher res file (24bit) from Buckingham & Nicks.
Crying In The Night (download FLAC)
- Dynamic Range (R128)= 3.9 LU
- Dynamic Range (DR) = 13
- Your expression = -7.6 dB
Not sure what all the numbers mean, but there's some obvious differences with the higher bit rate. Problem being, my library is all Redbook or album rips to Redbook with very few exceptions.
Volume Leveling should not require these controls.
And ya ... I agree, volume leveling SHOULD not require these controls, but hey ... just sayin'. After the R128 analysis, I ended up having to turn the amp up a couple notches, the DAC all the way, and the MC equalizer up +7 (otherwise flat) to get me back where I was for overall volume levels. I'd hate to think of what would happen to my speakers if someone were to inadvertently click adaptive leveling on ... it wouldn't be pretty. Probably scare the heck out of an astronaut on ISS who happened to be looking out the cupola when my cones when whizzing by ...
-
I ended up having to turn the amp up a couple notches, the DAC all the way, and the MC equalizer up +7 (otherwise flat) to get me back where I was for overall volume levels.
It is supposed to be a lot quieter than ReplayGain. One of the main reasons ReplayGain did not sound even, is because it did not provide enough headroom to adequately level all tracks.
If you're turning up the equalizer, you're going to cause playback to be uneven again. Turn up your amp.
Not sure what all the numbers mean, but there's some obvious differences with the higher bit rate. Problem being, my library is all Redbook or album rips to Redbook with very few exceptions.
It means that higher quality tracks need more headroom than low quality compressed releases. Sort your library by "Headroom" and see if you have anything approaching 0 dB.
Ideally you will have some tracks which are within a decibel or two of 0 dB, but none which exceed it.
-
In testing my new DAC I've noticed issues (digital noise) but just on some songs and I'm wondering if it is a combination of the downmixing of 6ch --> 2 ch content + the application of "Adaptive Volume" as it seems to only be this combination with the issue. Here is a screen shot that seems to show really bosted Peak Levels. I normally have it set to "Clip Protection" but changed it to "Flat line overflows" to see how much it overflows and on some of these tracks it hits 150%. That said, the digital noise is most noticeable in the quieter parts of the track where it is not overflowing.
Also happy for any suggestions, but turning off Adaptive Volume seems the go for now.
-
In testing my new DAC I've noticed issues (digital noise) but just on some songs and I'm wondering if it is a combination of the downmixing of 6ch --> 2 ch content + the application of "Adaptive Volume" as it seems to only be this combination with the issue. Here is a screen shot that seems to show really bosted Peak Levels. I normally have it set to "Clip Protection" but changed it to "Flat line overflows" to see how much it overflows and on some of these tracks it hits 150%. That said, the digital noise is most noticeable in the quieter parts of the track where it is not overflowing.
Also happy for any suggestions, but turning off Adaptive Volume seems the go for now.
One quick trouble shooting question: if you turn off both adaptive volume and volume levelling, does it show the same problem? I ask because the track you have highlighted has a clipping peak (i.e. +1dB). A +3dB overage would cause a 200% over, so a 1dB overage could conceivably cause a 20 to 40% overage at some points just from normal playback.
If there's no problem with both disabled, then it sounds like adaptive volume may not be interacting correctly with downmixing in this case.
-
Thanks - I need to play with this some more to work out exactly what I'm hearing on only a few of these 6ch tracks. Just wanted to check I had not missed something obvious with the DSP settings.
-
Got WASAPI Exclusive + Event Style Working which seems to have fixed the issue (but time will tell). More here (http://yabb.jriver.com/interact/index.php?topic=89230.msg613134#msg613134)
-
What are the steps to get to the above screen to make changes to Adaptive Volume?
-
Please see the first post in this thread.