Loudness management
- Lunkhead
- You're No Good
- Posts: 8175
- Joined: Sat Sep 25, 2004 12:14 pm
- Instruments: many
- Recording Method: cubase/mac/tascam4x4
- Submitting as: Berkeley Social Scene, Merisan, Tiny Robots
- Pronouns: he/him
- Location: Berkeley, CA
- Contact:
Loudness management
What would folks think about Song Fight joining the ranks of e.g. Apple Music and Spotify and publishing a guideline about loudness in submitted songs?
https://www.masteringthemix.com/blogs/l ... nd-youtube
More radically, maybe we could even explore doing some kind of loudness adjustment to the submitted songs to make for a smoother listening experience within a fight? I'm sure that would be extremely controversial though.
https://www.masteringthemix.com/blogs/l ... nd-youtube
More radically, maybe we could even explore doing some kind of loudness adjustment to the submitted songs to make for a smoother listening experience within a fight? I'm sure that would be extremely controversial though.
- fluffy
- Eruption
- Posts: 11097
- Joined: Sat Sep 25, 2004 10:56 am
- Instruments: sometimes
- Recording Method: Logic Pro X
- Submitting as: Sockpuppet
- Pronouns: she/they
- Location: Seattle-ish
- Contact:
Re: Loudness management
Isn't it already hard enough getting people to submit things with the right filename and format, and which are actually songs with the current title? Fortunately ReplayGain exists and there's plenty of tools out there to tell you what the LUFS is (heck, LAME does it by default, and a lot of mp3 encoders are based on LAME) so publishing a guideline is at least worth doing.
Regarding automatic normalization: Once upon a time I wrote a thing to stabilize volume and when word got out that I listened to all my music through it — especially Song Fight! — there was much wailing and gnashing of teeth because I was "destroying the musicians' intent." And when blue was looking into using it as a filter for the ongoing Song Fight! icecast stream, oh boy did people have a lot to say about that.
Hopefully the community has calmed down about that since then, because I think it's a good idea.
Maybe before the fight goes live the files could all be run through a (non-destructive) ReplayGain normalizer to target, say, -12dB.
Regarding automatic normalization: Once upon a time I wrote a thing to stabilize volume and when word got out that I listened to all my music through it — especially Song Fight! — there was much wailing and gnashing of teeth because I was "destroying the musicians' intent." And when blue was looking into using it as a filter for the ongoing Song Fight! icecast stream, oh boy did people have a lot to say about that.
Hopefully the community has calmed down about that since then, because I think it's a good idea.
Maybe before the fight goes live the files could all be run through a (non-destructive) ReplayGain normalizer to target, say, -12dB.
- grumpymike
- Push Comes to Shove
- Posts: 421
- Joined: Sun Apr 17, 2016 3:12 pm
- Instruments: Rage and curmudgeonry
- Recording Method: Zero-turn lawnmower
- Submitting as: Grumpy Mike
Re: Loudness management
You can do this non-destructively and automatically by providing a player that abides loudness metadata (Nur Ein site now does this - LUFS calculations are server-side and the client scales the volume slider) or by adding the ReplayGain and/or Apple tags to the MP3 files - which requires users to have capable players. They both have pros/cons but there is nothing to object to in either because they aren’t destructive. And they’re not mutually exclusive.
-
- Panama
- Posts: 978
- Joined: Mon May 15, 2017 1:29 pm
- Instruments: vox keys uke guitar
- Recording Method: REAPER
- Submitting as: Vowl Sounds, miscellaneous owl
- Pronouns: she/her
- Location: Madison, WI
- Contact:
Re: Loudness management
I don't think enforcing a loudness setting would be good unless there's something on the back end making it happen automagically, because it's a barrier to entry that I think would eliminate a lot of cool songs by non-tech-savvy musicians.
BUT I personally would like at least having a guideline to work to and help figuring out how to export my files at the right levels, because despite putting ozone and another LUFS meter in my chain on the master track (ps thanks to Chumpy and everyone else who chimed in on that Nur Ein Round Zero thread about measuring/fixing levels), I never seem to be able to export things at the right levels the "correct" way, for whatever reason--I have to futz around with a couple of different exports usually and the LUFS meter ends up saying it's too loud, but my ears end up saying it's fine next to reference tracks.
BUT I personally would like at least having a guideline to work to and help figuring out how to export my files at the right levels, because despite putting ozone and another LUFS meter in my chain on the master track (ps thanks to Chumpy and everyone else who chimed in on that Nur Ein Round Zero thread about measuring/fixing levels), I never seem to be able to export things at the right levels the "correct" way, for whatever reason--I have to futz around with a couple of different exports usually and the LUFS meter ends up saying it's too loud, but my ears end up saying it's fine next to reference tracks.
- Pigfarmer Jr
- Jump
- Posts: 2351
- Joined: Sat Mar 21, 2009 6:13 am
- Instruments: Guitar
- Recording Method: Br-900CD and Reaper to mix
- Submitting as: Pigfarmer Jr, Evil Grin, Pork Producer, Gilmore Lynette Tootle, T.C. Elliott
- Pronouns: he/him
- Location: Columbia, Missouri
- Contact:
Re: Loudness management
So I'm fairly inconsistent, but I try to render at -14 LUFS (because I read that's spotify etc., playback level.) Having a guideline that we should aim for is a great idea. Having a playback mechanism that levels out all the songs is a great idea (especially since we can all just download the songs and listen however we want, anyway.) Enforcing a level would be impossible (unless you disqualify entries) and I personally think isn't a good idea. That being said, I've not been bothered by volume differences enough to go to the trouble of listening through any system that "normalizes" playback. I just use the volume knob if/when needed.
Evil Grin bandcamp - Evil Grin spotify
T.C. Elliott bandcamp - T.C. Elliott spotify
"PigFramer: Guy and guitar OF MY NIGHTMARES." - Blue Lang
T.C. Elliott bandcamp - T.C. Elliott spotify
"PigFramer: Guy and guitar OF MY NIGHTMARES." - Blue Lang
- ujnhunter
- Ice Cream Man
- Posts: 1827
- Joined: Fri Mar 07, 2008 1:09 pm
- Instruments: Bass, Keyboards, Crummy Guitar & Animal Noises (especially Donkeys)
- Recording Method: Reaper 5.9x, Tascam FireOne/Behringer UMC202HD/Avid Eleven Rack/Line 6 UX2, Win 7 PC / Win 10 Laptop
- Submitting as: Cock, Chth*.*, D.A.H. (Der Alter Hahn)
- Pronouns: His Infernal Majesty
- Location: CT, USA
- Contact:
Re: Loudness management
Yeah, I wouldn't be a fan of Song Fight! altering the submitted songs that people download... If you wanted to make a Song Fight! "player" that did it the way that @grumpymike says that the Nur Ein player does, I wouldn't object to that... but altering a file that someone submitted wouldn't be cool in my book, and I, just like Stuntman Mike (no relation to @grumpymike), actually have a book.
P.S. How many commas is too many commas? Maybe I should just stick with my horrible... habit.
P.S. How many commas is too many commas? Maybe I should just stick with my horrible... habit.
-Ujn Hunter
Photovoltaik - Free 6 Track EP - Song Fight! Liner Notes
Photovoltaik - Free 6 Track EP - Song Fight! Liner Notes
Billy's Little Trip wrote:I must have this....in my mouth.....now.
- fluffy
- Eruption
- Posts: 11097
- Joined: Sat Sep 25, 2004 10:56 am
- Instruments: sometimes
- Recording Method: Logic Pro X
- Submitting as: Sockpuppet
- Pronouns: she/they
- Location: Seattle-ish
- Contact:
Re: Loudness management
I think the only alteration of the actual files that anyone's suggesting is adding a ReplayGain tag, which doesn't alter the audio data at all. It just tells the player where to set the volume knob for the song.
When I was using the volume stabilizer it was only as an audio filter that happened after my player (and it only turns up the volume on things that are too quiet and is careful to never cause clipping), and people got the misconception I was, like, changing the files or something. There's plenty of ways of dealing with automatic volume management that don't involve modifying the actual sound data, and remember that these things are all being designed and implemented by other musicians who understand the importance of protecting the recordings themselves.
When I was using the volume stabilizer it was only as an audio filter that happened after my player (and it only turns up the volume on things that are too quiet and is careful to never cause clipping), and people got the misconception I was, like, changing the files or something. There's plenty of ways of dealing with automatic volume management that don't involve modifying the actual sound data, and remember that these things are all being designed and implemented by other musicians who understand the importance of protecting the recordings themselves.
- Rabid Garfunkel
- Jump
- Posts: 2468
- Joined: Sat Sep 25, 2004 12:43 pm
- Instruments: Absurdity
- Recording Method: iPhone, GarageBand & rando apps/toys
- Submitting as: OZYMANDIPUS, Rabid Garfunkel, Primitive Screwheads
- Pronouns: that guy
- Location: Portland, Oregon
- Contact:
Re: Loudness management
Could this be an addition to the jukebox's playback function, leaving the front page fight links as submitted?
Oh, and hey all. Been a while, yo.
Oh, and hey all. Been a while, yo.
- jast
- Ice Cream Man
- Posts: 1326
- Joined: Tue Jul 29, 2008 7:03 pm
- Instruments: Vocals, guitar
- Recording Method: Cubase, Steinberg UR44
- Submitting as: Jan Krueger
- Pronouns: .
- Location: near Aachen, Germany
- Contact:
Re: Loudness management
I don't mind an official recommendation, though I doubt they'll make much of a difference.
I don't mind an optional filter in e.g. the jukebox, either.
I wouldn't be happy with any automatic mangling of audio - for one thing it can destroy the intent of the mix, for another it makes it hard to review the mix if you don't know what's due to the original submission and what's due to filtering. I prefer listening to the originals.
Finally, I don't really think ReplayGain tags are really worth the effort here. As I understand it, ReplayGain calculates the recommended adjustment based on RMS normalization and an extra 14 dB of headroom. However, this fails if a track's dynamics exceed 14 dB above the RMS mean - the adjustment will either have to include more headroom for this track, or cause clipping. So, we either get different loudness in different tracks, or artifacts.
This might work if all of the tracks actually are mastered to similar specifications, but in practice they won't be - and it's unrealistic to ask this of participants who aren't really that much into the engineering side of things.
I don't mind an optional filter in e.g. the jukebox, either.
I wouldn't be happy with any automatic mangling of audio - for one thing it can destroy the intent of the mix, for another it makes it hard to review the mix if you don't know what's due to the original submission and what's due to filtering. I prefer listening to the originals.
Finally, I don't really think ReplayGain tags are really worth the effort here. As I understand it, ReplayGain calculates the recommended adjustment based on RMS normalization and an extra 14 dB of headroom. However, this fails if a track's dynamics exceed 14 dB above the RMS mean - the adjustment will either have to include more headroom for this track, or cause clipping. So, we either get different loudness in different tracks, or artifacts.
This might work if all of the tracks actually are mastered to similar specifications, but in practice they won't be - and it's unrealistic to ask this of participants who aren't really that much into the engineering side of things.
- grumpymike
- Push Comes to Shove
- Posts: 421
- Joined: Sun Apr 17, 2016 3:12 pm
- Instruments: Rage and curmudgeonry
- Recording Method: Zero-turn lawnmower
- Submitting as: Grumpy Mike
Re: Loudness management
While it might not be in the spirit of the ReplayGain standard, you can calculate the value in REPLAYGAIN_TRACK_GAIN however you like. The same is true for iTunNORM. So, you could calculate and adjust the loudness using LUFS in order to be more accurate in terms of perceived loudness. Regardless, you would still need a target, which likely entails clipping. Not to re-litigate the entire concept of standardizing loudness, but the reality is people expect songs to be playable adjacent to other songs.
The larger problem is that few audio players support these tags.
- fluffy
- Eruption
- Posts: 11097
- Joined: Sat Sep 25, 2004 10:56 am
- Instruments: sometimes
- Recording Method: Logic Pro X
- Submitting as: Sockpuppet
- Pronouns: she/they
- Location: Seattle-ish
- Contact:
Re: Loudness management
Or if they do support the tags it's a very all-or-nothing thing, and is often shoddily-done (like iTunes).
I wonder how many entries in the archive already have ReplayGain tags without anyone even realizing it.
I wonder how many entries in the archive already have ReplayGain tags without anyone even realizing it.
- Lunkhead
- You're No Good
- Posts: 8175
- Joined: Sat Sep 25, 2004 12:14 pm
- Instruments: many
- Recording Method: cubase/mac/tascam4x4
- Submitting as: Berkeley Social Scene, Merisan, Tiny Robots
- Pronouns: he/him
- Location: Berkeley, CA
- Contact:
Re: Loudness management
I personally was not proposing any modification of the audio in the files, just adding metadata. So could we maybe please stop wasting time arguing about something that wasn't mooted in the first place?
- jb
- Hot for Teacher
- Posts: 4165
- Joined: Sat Sep 25, 2004 10:12 am
- Instruments: Guitar, Cello, Keys, Uke, Vox, Perc
- Recording Method: Logic X
- Submitting as: The John Benjamin Band
- Pronouns: he/him
- Location: WASHINGTON, DC
- Contact:
Re: Loudness management
It's not compression or EQ, it's gain. It doesn't seem like normalizing to a particular LUFS number would destroy the intent of the mix on Song Fight any more than it does on Spotify.
I can see your point for any production that is intentionally trying to screw with the listener, experimenting with extreme quietness or extreme loudness, or farcical submissions like a Lightning Ear Fart. But for %99.999 of submissions this seems like it would at worst do no harm and at best help them be heard in the best context possible.
If someone suggested automatically compressing or EQ'ing the entries, I think I would probably agree with you. But this is just about gain, so given an option to listen to a LUFS-normalized playlist or the original submissions, I'd choose the normalized playlist to make it easier for me to listen to the songs confidently without having to adjust my volume controls for each song.for another it makes it hard to review the mix if you don't know what's due to the original submission and what's due to filtering. I prefer listening to the originals.
We make fun of this idea, but this really is Song Fight, not Production Fight. So in my opinion, someone's lack of engineering expertise should not disqualify them from winning a fight. It's a fact that humans confuse loudness for fidelity, so for submissions that are too quiet increasing the LUFS level via a normalization process would be leveling the playing field to some extent. And for submissions that are too loud, it can prevent the listener from skipping the song entirely just because it came on too loudly after the previous song.
I do think this probably wouldn't work well unless the processing happened on the server-- I think player metadata and implementation might be too much of a crapshoot given the various methods people use to listen to Song Fight (the variety of which always surprises me).
Just my $.02
JB
blippity blop ya don’t stop heyyyyyyyyy
- grumpymike
- Push Comes to Shove
- Posts: 421
- Joined: Sun Apr 17, 2016 3:12 pm
- Instruments: Rage and curmudgeonry
- Recording Method: Zero-turn lawnmower
- Submitting as: Grumpy Mike
Re: Loudness management
If you just want to post guidelines, do it. I don't think anyone will object.
- fluffy
- Eruption
- Posts: 11097
- Joined: Sat Sep 25, 2004 10:56 am
- Instruments: sometimes
- Recording Method: Logic Pro X
- Submitting as: Sockpuppet
- Pronouns: she/they
- Location: Seattle-ish
- Contact:
Re: Loudness management
Also keep in mind that at least for ReplayGain it's only choosing one relative gain offset for the entire track, meaning that if someone's made a purposefully very quiet VERY LOUD!!!! mix (for example, Octothorpe's "Crinkle Binkle") this wouldn't do anything to affect that.jb wrote: ↑Tue Aug 27, 2019 11:45 amI can see your point for any production that is intentionally trying to screw with the listener, experimenting with extreme quietness or extreme loudness, or farcical submissions like a Lightning Ear Fart. But for %99.999 of submissions this seems like it would at worst do no harm and at best help them be heard in the best context possible.
(Also tracks like that tend to be very listener-hostile and not really all that funny)
- Lunkhead
- You're No Good
- Posts: 8175
- Joined: Sat Sep 25, 2004 12:14 pm
- Instruments: many
- Recording Method: cubase/mac/tascam4x4
- Submitting as: Berkeley Social Scene, Merisan, Tiny Robots
- Pronouns: he/him
- Location: Berkeley, CA
- Contact:
Re: Loudness management
Spotify's FAQs on the subject for reference:
https://artists.spotify.com/faq/mastering-and-loudness
Turns out at playback time in mobile apps they apply a limiter when boosting softer tracks:
https://artists.spotify.com/faq/mastering-and-loudness
Turns out at playback time in mobile apps they apply a limiter when boosting softer tracks:
Positive gain is applied to softer masters so that the loudness level is at ca - 14 dB LUFS. A limiter is also applied, set to engage at -1 dB (sample values), with a 5 ms attack time and a 100 ms decay time. This will prevent any distortion or clipping from soft but dynamic tracks.
- fluffy
- Eruption
- Posts: 11097
- Joined: Sat Sep 25, 2004 10:56 am
- Instruments: sometimes
- Recording Method: Logic Pro X
- Submitting as: Sockpuppet
- Pronouns: she/they
- Location: Seattle-ish
- Contact:
Re: Loudness management
Previously I was determining my LUFS by just looking at what the ReplayGain output said from my mp3 encoding, which seems to not be a great way of doing it. So just now I found a free VST/AU that looks like it'll give you a better overview of what's happening dynamics-wise.
- jast
- Ice Cream Man
- Posts: 1326
- Joined: Tue Jul 29, 2008 7:03 pm
- Instruments: Vocals, guitar
- Recording Method: Cubase, Steinberg UR44
- Submitting as: Jan Krueger
- Pronouns: .
- Location: near Aachen, Germany
- Contact:
Re: Loudness management
Yeah, I was using Youlean's for quite some time, too. I've switched to a DAW that has built-in visualization of integrated loudness, though, so that makes it unnecessary.
- jast
- Ice Cream Man
- Posts: 1326
- Joined: Tue Jul 29, 2008 7:03 pm
- Instruments: Vocals, guitar
- Recording Method: Cubase, Steinberg UR44
- Submitting as: Jan Krueger
- Pronouns: .
- Location: near Aachen, Germany
- Contact:
Re: Loudness management
Yeah, same for REAPER... anyway, back to the topic.
I think that normalizing loudness is a particularly tricky topic for submissions that have a really rough mix - e.g. the drums kick in late and are way overbearing, stuff like that. Inexperience can easily make for fluctuations of 6 dB, which would mean a departure of maybe 3 dB in the mean - so, part of the song is 3 dB louder than the rest, and the rest is 3 dB quieter. The same can happen for a song with two distinct parts - in a more dynamic mix, 6 dB of difference are not at all out of the question (and not as extreme as the examples brought up earlier). There's really no way to stop this without going into audio-level compression (which we seem to agree is not a good idea).
So, we're likely going to have 3-4 dB variation in fights even with normalization tags provided - and in general I think most songs don't stray that much further - this is pretty much mainly about songs that are mastered way too hot (at least in my opinion). At that point I have to wonder: is this enough of a pain point to justify setting up extra processing infrastructure, or add more manual steps to publishing each fight?
(For me personally it's not...)
I think that normalizing loudness is a particularly tricky topic for submissions that have a really rough mix - e.g. the drums kick in late and are way overbearing, stuff like that. Inexperience can easily make for fluctuations of 6 dB, which would mean a departure of maybe 3 dB in the mean - so, part of the song is 3 dB louder than the rest, and the rest is 3 dB quieter. The same can happen for a song with two distinct parts - in a more dynamic mix, 6 dB of difference are not at all out of the question (and not as extreme as the examples brought up earlier). There's really no way to stop this without going into audio-level compression (which we seem to agree is not a good idea).
So, we're likely going to have 3-4 dB variation in fights even with normalization tags provided - and in general I think most songs don't stray that much further - this is pretty much mainly about songs that are mastered way too hot (at least in my opinion). At that point I have to wonder: is this enough of a pain point to justify setting up extra processing infrastructure, or add more manual steps to publishing each fight?
(For me personally it's not...)
- Pigfarmer Jr
- Jump
- Posts: 2351
- Joined: Sat Mar 21, 2009 6:13 am
- Instruments: Guitar
- Recording Method: Br-900CD and Reaper to mix
- Submitting as: Pigfarmer Jr, Evil Grin, Pork Producer, Gilmore Lynette Tootle, T.C. Elliott
- Pronouns: he/him
- Location: Columbia, Missouri
- Contact:
Re: Loudness management
To further the quote from Lunkhead:
Also, we have the option of setting up something we could do individually to help with the listening experience. If someone has a player/normalization process that each of us could individually set up/use, then it's an option I'd probably choose to use. For instance, if the jukebox was set up in such a way that it had the option or was standard with normalization then I'd use that even more than I already do. (I'd really like it to be optional to hear the differences, but I'm not sure that's necessary.)
I've always erred on the side of softer, ie - 14.2 etc., but according to this, it'd be better to be at -13.8 so that no limiting is put on the track. Although, I doubt I could hear the difference. It'd be interesting to do a listening test, but not worth the trouble.Negative gain is applied to louder masters so the loudness level is at ca - 14 dB LUFS. This process only decreases the volume in comparison to the master; no additional distortion occurs.
Positive gain is applied to softer masters so that the loudness level is at ca - 14 dB LUFS. A limiter is also applied, set to engage at -1 dB (sample values), with a 5 ms attack time and a 100 ms decay time. This will prevent any distortion or clipping from soft but dynamic tracks.
Also, we have the option of setting up something we could do individually to help with the listening experience. If someone has a player/normalization process that each of us could individually set up/use, then it's an option I'd probably choose to use. For instance, if the jukebox was set up in such a way that it had the option or was standard with normalization then I'd use that even more than I already do. (I'd really like it to be optional to hear the differences, but I'm not sure that's necessary.)
Evil Grin bandcamp - Evil Grin spotify
T.C. Elliott bandcamp - T.C. Elliott spotify
"PigFramer: Guy and guitar OF MY NIGHTMARES." - Blue Lang
T.C. Elliott bandcamp - T.C. Elliott spotify
"PigFramer: Guy and guitar OF MY NIGHTMARES." - Blue Lang
- jb
- Hot for Teacher
- Posts: 4165
- Joined: Sat Sep 25, 2004 10:12 am
- Instruments: Guitar, Cello, Keys, Uke, Vox, Perc
- Recording Method: Logic X
- Submitting as: The John Benjamin Band
- Pronouns: he/him
- Location: WASHINGTON, DC
- Contact:
Re: Loudness management
Kind of an "advanced mode" sort of thing.Pigfarmer Jr wrote: ↑Wed Aug 28, 2019 5:13 amAlso, we have the option of setting up something we could do individually to help with the listening experience. If someone has a player/normalization process that each of us could individually set up/use, then it's an option I'd probably choose to use. For instance, if the jukebox was set up in such a way that it had the option or was standard with normalization then I'd use that even more than I already do. (I'd really like it to be optional to hear the differences, but I'm not sure that's necessary.)
Personally I lean towards the "how can we make all submitted productions as good as possible considering the variety of engineering expertise"-- so for example "Hey LANDR want to partner up and give any songfight who wants it the option to pass their submission through your maxifier gizmo?" An opt-in auto-master for those who are either lazy or inexperienced.
Because 15-16 Puzzle. Because 2002 JBB.
But that's just me.
J
blippity blop ya don’t stop heyyyyyyyyy