Hacker News new | past | comments | ask | show | jobs | submit login
Loudness war (wikipedia.org)
272 points by luketheobscure on Jan 7, 2020 | hide | past | favorite | 172 comments



The 20,000 Hz podcast provided a full history of this, discussing various samples along the way:

https://www.20k.org/episodes/loudnesswars

It's pretty standard for that podcast, which features lots of stories about sound and sound engineering. There's another interesting article on creating the most silent possible room, and how eerie it feels to be in it. https://www.20k.org/episodes/silence

Another on how restaurants got louder and louder over time, that deep bass sound that's taken over film, another on scientific experiments trying to measure if the Stradivarius is as great a violin as everyone insists, another on the tangled history of the iconic Price is Right theme song... so many really.

Great podcast, lots of range given the topic.


This sounds interesting and I will listen to it, but for a podcast about sound engineering, I really resent that their website doesn't provide an option to download the bare mp3/m4a. I had to use Inspector.

No subscribe via RSS option either, only Apple Podcasts and Google Podcasts. This is becoming depressingly common.


It does, but it's buried in the player.

Download: In the player click "share" and then click the "down arrow" icon.

RSS feed: In the player click "subscribe" and then click the RSS icon (it's the first one)


Thank you. Wow. Why the heck would you put the download link in the Share menu?!?


This is a standard iOS pattern. Agreed that it's all but intuitive but that's how iOS is these days.


I see where you're coming from, but to me that's a bit different.

1. iOS heavily deemphasizes the concept of a filesystem. So when I download something, what I'm really doing is "sharing" that media with another app.

2. I associate iOS's box-and-arrow icon with more than just "sharing". This player doesn't have an icon, you literally have to click text that reads "share".

It's perfectly possible the people who designed this player thought they were referencing iOS, but IMO they very much missed the mark. (Not to imply that iOS's method is particularly good UX either!)


That makes sense, I’m with you. But I always felt that Share was sufficient for any externally targeted location. I’ll have to look at this thing, though. Haha.


And, to complete the circle, the iOS way is the audio engineers like. Even these days, audio engineers are tied to Apple - definitely macOS, but increasingly iOS as well.


Pro Tools is a large part of the industry market and that’s multi-platform.

What do you mean by “audio engineers are tied to apple”?


Not sure if it's still the case but for many years Apple's Core Audio was a big draw for people running audio software due to it's stability, performance and compatibility with various devices. I've used Windows with ASIO drivers for many years and always found it a bit buggy with occasional random drop-outs that can only be fixed by reinstalling the drivers. Pulse Audio on Linux can be even more of a nightmare to get working with your equipment, but I've heard it's improving recently.


That’s a good point. In my experience in studios (from around 5 years ago) windows is just about as prevalent as OS X at this point. I’m sure whatever niche you’re in does affect this however.


They are not tied to apple, but it is often the preferred OS.

I haven't used MacOS/OSX in a couple years, but at least back then it was in my opinion by miles the best out of the box experience for audio related things.

E.g it took me hours on Windows 10 to have a basic setup with reliably low latency, without clicks, noises, dropouts and generally getting the OS out of my way (e.g. automatic rebooting for updates is kinda suboptimal if you plan to use your laptop for live music). Never had anywhere near the same level of issues on macs.

Sure you can make it work on Windows or Linux, but if this were my main job, I'd need very specific reasons or strong personal os preferences outside audio to not pick an Apple.


Movie composers are often Windows, not Mac. Even after the new MacPro It's much easier and cheaper to put together a monster PC with no-compromise specs for loading huge sample libraries and running huge mixes.

Also, these people have people to do tech support, setup, and configuration for them - often onsite - so mostly their systems "just work."


Apple had the money to buy "Logic" from a German company called "Emagic", and scrap their Windows version. Logic then soon became industry standard.

Voilà.


Logic isn't the industry standard though. PT has long been thought of as the industry standard, although that's changed recently with Cubase/Ableton Live/FL/Logic becoming massively popular as well. Really depends on genre/application (EDM, acoustic, film/game/media, etc).


It's way more likely to be ableton, reaper or logic now.


It's very dependent on application niche. Movie and game composers are more likely to use Cubase or Nuendo. Very big pop studios (includes rock, country, rap, etc) are likely to use ProTools. Smaller home studios are likely to use Logic or Cubase. EDM artists may be using Ableton Live, Reaper, Logic, BitWig, Cubase, or sometimes ProTools.

There really is no "industry standard" now. PT was the standard maybe ten years ago, and still has traction in studios where artists need to move their sessions to another facility. But Avid pissed off everyone by overcharging for buggy software, so there's a lot more diversity now.


> Avid pissed off everyone by overcharging for buggy software, so there's a lot more diversity now.

Huh, it will be interesting to see if Adobe ends up in a similar position ten years from now as a result of their pricing strategy. In the graphic design space, which I'm more familiar with, it feels like there's very slow but still real migration away from Creative Suite.


StudioOne has gained a ton of popularity in home and small studios over the past half decade as well.

The DAW market has very healthy competition right now!


Logic on macOS is the pro-audio industry standard.


No, it's not.

The only thing in the pro audio industry that is close to being an industry "standard" is the continued use of Pro Tools as the preferred tracking software. Even that is starting to fade outside of the coastal cities.


Part of this I think is related to very low latency on iOS. IIRC Android had severe problems with it. Maybe still has?


Yes, OSX and iOS both perform better than Android when it comes to round-trip latency. Core Audio in general was a real godsend to audio folks for many years, and was a compelling reason to stay in the Apple ecosystem.

That relationship and trust has soured considerably since then, but until there's a new "Apple" in town for audio folks, I don't invision ship-hopping for quite a while. It is an utter disgrace getting Windows 10 to work well as a studio PC, and most audio engineers simply won't or can't use the vastly superior FOSS pro audio software out there today.


Also, Perfecting Sound Forever (book) by Greg Milner. I believe he was interviewed for that podcast. It’s a truly fantastic read which goes in depth in the history of recorded sound and has a ton of information on the loudness wars.


That episode in particular was so outstanding. They really did a great job of mixing in the samples to make you hear what they were talking about. I loved the Stradivarius one too! They’ve been knocking it out of the park lately.


It's a great podcast! I can also recommend the episode on the THX sound.

https://www.20k.org/episodes/thxdeepnote


Came here to post this. One of my favorite podcasts! Highly recommended


The Loudness War is, thankfully, starting to taper off - largely because of YouTube ( https://productionadvice.co.uk/youtube-loudness/ ) and Spotify ( https://artists.spotify.com/faq/mastering-and-loudness#what-... )

Since YT and Spotify are some of the predominant ways people listen to music these days, and they normalize loudness, music producers are starting to go back to more normal masters, thankfully.


I've always been curious about Spotify's approach on this, because truth be told the service on good quality speakers does sound rather good.

I'm very pleased by everything I just read in this link — they really think of the user first, and favor music quality and the general experience. It's exactly how I did/do it myself, manually.

Awesome, really awesome to learn that the loudness war is finally coming to a close.


I don't know how normalization is done, but I do know that Spotify are very professional when it comes to audio quality. They do double-blind testing with setups by audio engineers to evaluate end-to-end quality when evaluating changes.


The double-blind scientific approach to audio testing ("ABX") has long been a major point of contention between "objectivists" and "subjectivists"[1]. This has been a godsend for the marketing marketing of some snake-oil companies in audiophile circles.

Really glad to know Spotify is on the side of science, here.

[1]: This blog has got to be the most interesting I ever read on whatever topic it touches, including the objective vs subjective debate: http://nwavguy.blogspot.com/2011/05/subjective-vs-objective-...


Do you mean double-blind testing benefits snake-oil companies? If so, may you expound on that? To my understanding, double-blind tests are meant to avoid un/conscious bias during testing.


No, on the contrary! Sorry for my ambiguous writing. Indeed double-blind is the only way to go.

What I meant is that there is a sizeable chunk of the audiophile industry (a certain press, electronics brands, stores and even studios themselves) that conveniently avoids any and all scientific testing, and promotes typical snake-oil "features" and "specs". Quite sadly for their abused customers.

One of the worst trends in my opinion was faking technical format quality by using different masters on each — with the shitty master on CD/MP3 and the good one, more dynamic, on DVD/SACD/vinyl etc. The latter always sounded better simply because it wasn't the same source!

In such matters, I guess we can trust Spotify based on their claims, and anecdotally I tend to believe them.


I assume they set a target of, say, -14 LUFS (Loudness Units Full Scale), take a reading of the integrated value for the whole track, and then turn the volume of the whole track down by whatever dB value will hit the target.

This avoids the problem of cruder methods, where you take, say, the peak dB as the "loudness" of the track. Doing it that way would take a fairly quiet song with one overly dramatic drum hit and make it so you couldn't hear the rest.

EDIT: I think they will also turn a track up if it's under the target, but I'm not absolutely sure. There would be no real reason not to.


I think they mentioned ReplayGain with AAC and ogg formats, but it's possible they use some custom implementation. There's a good deal of information[1]; it is indeed not peak normalization:

> ReplayGain is different from peak normalization. Peak normalization merely ensures that the peak amplitude reaches a certain level. This does not ensure equal loudness. The ReplayGain technique measures the effective power of the waveform (i.e. the RMS power after applying an "equal loudness contour"), and then adjusts the amplitude of the waveform accordingly. The result is that Replay Gained waveforms are usually more uniformly amplified than peak-normalized waveforms.

Generally we indeed target both up or down in loudness (although 'up' is quite rare, typically high-end recordings of classical or jazz or otherwise very dynamic pieces).

Honestly, everything I read screams that Spotify really hired the right people to make these decisions.

Now I only have to rent about their handling (or rather, lackthereof) of metadata — how hard is it to display an actual first release date? but meh. Audio quality is great, no question about it. I just wish they'd stream FLAC in very-high quality + "quiet" environment settings (the "pure hi-fi" experience so to speak), but that'll come in due time I suppose.

[1]: https://wiki.hydrogenaud.io/index.php?title=ReplayGain


YouTube's volume normalization seems to only reduce the volume of overly loud video, but doesn't add gain to videos that are too quiet.

I guess they don't want to bother with limiters, so understandable.


Musicians and engineers don't want their work 'modified' by the streaming service to make it louder.


I was hopeful when I heard about mainstream loudness normalisation. I've been using ReplayGain for years and it's still superior due to storing both track and album level. It's unclear to me what Spotify etc is doing, but it's is track normalisation it's going to ruin albums.

There's a bigger problem, though. My girlfriend complained about volume fluctuations in her car between tracks. She uses an iPhone so I enabled "Soundcheck" (Apple's normalisation). It works fine in the car. But now she complains that headphones simply aren't loud enough. It turns out that the volume controls are essentially made for overly-compressed, loud content and can't sufficiently amplify dynamic content.

So, unfortunately, people will just turn it off and there's still an advantage to mastering "loud".


In my experience, noise-cancelling headphones have made a big difference. The music doesn't need to compete anymore and you can lower the volume and enjoy the dynamic range.

I also wonder if the resurgence of vinyl, and rise of home producers on bandcamp/soundcloud/etc have helped contribute.


Noise cancelling headphones and ear plugs have been a god send. I just recently got the new AirPods and the noise cancelling in those is phenomenal, I've only experienced better in bulky over ear headphones. If you're in the market for noise cancelling ear plugs, I can definitely recommend them. The mic doesn't seem very good though, getting a lot of complaints about low volume. Oh well, can't have it all!


Does this discourage folks from applying excessive dynamic range compression? Compressed tracks still sound louder.


With normalization, a track with excessive compression and a normally mastered track will have the same loudness. The compressed one will simply have less dynamic range.


It reduces the appeal, but compressed tracks will still sound louder. It depends how advanced the normalisation is. The ideal situation would be something like Dolby's dialogue normalisation where the level is determined from a specific part of the soundtrack rather than the entire thing. It's not obvious how this would work with music, though.


The volume normalisation on Spotify is optional. I can't remember if it's on or not by default.


From what I recall, it's on and set to 'medium' by default. It's become somewhat of a habit to set it to 'quiet' any time Spotify shows me a release announcement (which tells me the settings got reset to default)


that link does not show at all how spotify's policy counters loudness wars. it actually says that they apply a limiter the same as you would to get a loud master. so essentially if a musician avoids the loudness war, spotify applies its principles no matter what


I think you're misunderstanding what the issue is with the loudness war. The problem isn't making a loud master - it's that the dynamics are crushed and the audio signal clips.

Spotify normalizes volume to a LUFS level, which is not nearly the same thing as overly compressing dynamics and causing clipping.

https://www.sweetwater.com/insync/what-is-lufs-and-why-shoul...


AFAIK they simply mess up your mastering if you don't comply.


Where I was born it used to be dead silence at night till about a decade ago.

Then the buses started announcing station names outloud. As the announcement is loud and the bus windows are open, I get woken up from it almost every night.

Next street lights started to beep for blind people. While this is a great idea, in practice the level of sound they set makes sense for the day (when the street at busy), but for the night it is way too strong, going into people houses. I got used to sleep with a constant weak beeping sound.

Then the e-scooters came, and created three sources of new noise during the night. One when someone touches them ("stealing alert!"), one when someone is looking for them ("location beep") and the last when the van of the company comes to pick them up around 3-4 am.

sigh, I'm moving to a cabin in the desert.


The article is about applying dynamic range compression and similiar techiques to increase audio levels in recorded music.


I know, I took the liberty of sharing in the comments a story of increasing audio level in society in general as I think these are connected social issues.


I agree, nice comment.

Recently stayed in a recently modernized hotel room. Lots of beeps...


I totally get what you're saying, but there's some contradiction which I don't understand: you say a decade ago it was 'dead silence' but also that it ended when 'busses started announcing' which seems to imply the busses were there already before. So with 'dead silence' do you mean 'all the usual traffic noise' or ... ?


I think I tend to forgive noise that isn’t abrupt, gradually increases /decreases in loudness, and isn’t high pitched, and never gets too loud.


The Roman philosopher Seneca made a similar point about non-continuous noise.

"Among the noises that sound around me but do not distract me, I count passing carriages, a carpenter somewhere in the building, a nearby saw grinder, and that fellow who demonstrates flutes and trumpets near the Meta Sudans, not so much playing them as bellowing. Even now I find noises that recur at intervals more bothersome than a continuous drone."

You can find more of his thoughts on the matter including his wonderful description of living above a bathhouse in Letter 56 to Lucilius.


That’s not strange at all. Most of the new noises you listed were specifically designed to catch your attention, while motor vehicles are constructed to be as quiet as it’s possible.


>while motor vehicles are constructed to be as quiet as it’s possible.

And before anyone sails on in to try and win some free virtue points about how OEMs slap excessively loud exhausts on stuff because buyers prefer that I would just like to point out that we have had ever tightening standards for idle, drive-by and wide open acceleration noise a vehicle can make since the 1980s.

Buyers generally want their cars to make "some" discernible change in engine noise when they floor it but that's about it. Excessive noise is fatiguing whether people realize it or not and OEMs likewise try to avoid it because it makes ownership less pleasant. For this reason when they do slap "performance" exhausts on their sportier models it's a case of "how quiet can we get away with" (and many automakers use resonance tubes, engine sounds over speakers and other tricks in order to be quiet yet still provide noise when the driver floors it). The people designing tire treads, mirrors, the wheel well opening all have reducing noise as a priority. Reducing noise makes for an all around less fatiguing experience and has really turned into an arms race among the OEMs since the mid '00s or so.


> while motor vehicles are constructed to be as quiet as it’s possible.

The interior, perhaps.

I live on a street which has a relatively low limit (30 mph) but cars accelerate as if they were about to join a freeway...they are not quiet at all.


Urban areas are going to be so much better once EVs are dominant.

Unfortunately I live near a highway, and the dominant noise isn't engine noise but tire/wind noise from the cars.


You're right. There were cars passing and pedestrians, but their noise didn't penetrate into the apartment. By "dead silence" I meant what's being heard from inside the house.


In Sweden the street lights are rather tocking, which is much less disturbing, than beeping for the blind. And the sound level adapts to the surrounding noise level, so they are much less loud during the quiet of the night than during the busy hours.


Speaking of lights, are there no laws in the US about how bright headlights can be? At the risk of being too dramatic, I have had cars behind mine with their lights so bright I thought they had their high beams on.


That's the idiocy in general of modern car design.

A bright-blue headlight of a modern car will actually make everything surrounding the headlights darker. A light which is slightly dimmer and more shifted towards red works much better for your peripheral vision. If you drive in rural areas the difference is very apparent.

Not only that, but street lights seem to be doing that as well.

Cross-walks here are now illuminated along the path with a strong shaped light. But the light is so bright that during night they just blind the observer: the pedestrian looks like ghost in a black background. IMHO this is even more dangerous, I frequently cannot see past the crosswalk, so a pedestrian which is passing behind it is risking much more than before. Go figure.

There are a couple of intersections which I pass frequently where the green light is too bright already during day. During the night, as soon as you get the green light you get blinded, which is _awesome_ since the light is guarding a cross-walk in this case as well. By night you cannot see pedestrians when you have the green.


The police are the biggest offenders when it comes to unnecessarily blinding drivers at night.

The construction companies (who have to pay their own insurance premiums and generally avoid unnecessarily risking injury to people by blinding drivers) have long since toned down the lights they used (there was a short time period where they all had super bright lights because LEDs were new and cool so why not have a 1000W flasher), switched to the non-glaring light plants that use the canvas bags.


Ok I know this is totally off-topic, but recently it indeed seems as if something changed (and I'm in Europe). Either the newer LED (or whatever they use in cars) are brighter, or they are aimed differently (that makes a lot of difference), or the rules loosened, or a combination of those, or something else which I can't figure out. It's not normal to be blinded when looking in your rear mirror, right?

Slightly related, still widely off-topic: lights on bicycles, same problem. I think e.g. Germany has rules for those, but in countries where there aren't and/or there's zero awareness being raised it's sad to see how many people ride around with LED headlights which are way stronger than required and aimed straight ahead instead of down, completely blinding opposite direction traffic including cars. Now on the commute I do I've made it a sport to (friendly) shout to them telling to put their light down and it seems to have effect. But there's still a long way to go.


Headlights got brighter (no new car uses anything like a classic bulb, going for Xenon or LED most of the time) and cars got a lot taller on average (SUVs are more popular than ever) so you're getting more light even when the car is further away behind you. I don't think regulation changed much regarding this.


Also on some of the cars with EU daytime running lamps, they put the LEDs really low, where only fog lights were once permitted (fog lamps were illegal to use when not foggy), presumably firing up or entirely non directional. These are surprisingly good at dazzling, especially when roads are wet. There's a few makes doing this, but a couple of models are particularly problematic.


All EU cars have DRL now. But the LEDs are basically what used to be the marker lights on any headlight (the first step in turning on the headlights). More commonly now they are overlapped with the signal lights so when signaling the LEDs just turn from white to yellow. But they're almost always still in the headlight block. When for some reason a manufacturer decides to put them in the fog lamps they run at reduced intensity. Just like when the fog light turn on towards the side you are cornering to help with more light.


> Speaking of lights, are there no laws in the US about how bright headlights can be?

See also bicyclists.

There is a general saying that there are two types of lighting: those that allow you to see, and those that allow to be seen. In most urban areas with street lamps, one simply needs to be seen. Too many of my fellow pedallers do not seem to understand this distinction, and do not realize that more lumens does not always been better: at some point you're simply blinding people, and they can no longer tell where you are.

A moderately (500 lm?) bright blinking light, with a simple on-off, non-random pattern seems to be best IMHO.


There are laws, but it doesn't mean they are properly enforced.

Chances are that the car behind you had misaligned headlights or some dubious aftermarket modification, like a HID conversion kit.


Most vehicles will be born and die bone stock. The internet greatly over-estimates the number of vehicles that get aftermarket lights (LED dome light replacement bulbs from Autozone notwithstanding).


Badly adjusted headlights are very common though.

You are supposed to adjust them based on the weight distribution, but people tend to use the highest setting, because that's the best for them. But if they have a lot of weight on rear, the beams are too high and can dazzle other drivers.

And sometimes the optics are simply misaligned because of shock, wear, vibration, ... even dirt can change the beam shape and be problematic. This is especially true for HID lamps, which are closer to a point light source and require higher precision optics compared to halogen light bulbs. And because it mostly affect others, these problems tend to go unnoticed by the driver.


There are, but they regulated wattage, not luminosity.

An old crummy incandescent bulb in a 90s civic will produce a lot less light at the same wattage, than a new, super-bright LED.


That's stupid. I think when these laws were made wattage was a good proxy for luminosity. It should be updated.


You're probably losing your stars from encroaching light pollution (and your sleep due to blue-heavy LED's) as well.


The noisy things you're complaining about are mostly an American phenomenon.


While I despise the loudness war and the loss of dynamic range, it's worth recognizing that where and how people listen to music has changed drastically in the last 20 years.

Moreso than ever in the past, people listen to music in public or otherwise noisy environments. (worse yet) They listen with generally low quality headphones/earbuds.

Basically, if the music isn't compressed, they won't hear most of it.

Of course, this plus the crappy earbuds and the noisy surroundings means they listen at higher volumes and suffer more hearing damage, which leads to less human sensitivity to sound levels (and more desire for compressed, high volume music).

It's very rare for most of us to get to sit in a good room with a good system and listen without interference... but it's really, really nice to do when you can.


I don't see why this needs to be done at the track mastering level though. If audio device manufacturers know their users are primarily going to be listening in noisy environments with cheap headphones they can compress the track at playtime.


As a person studying music production, this can't be done easily without having to mix and then master again. Certainly there could be varying edits of a track, though this is already common practice. The music you hear o the radio, or even Spotify, will usually be mastered differently than what you'll hear on a CD or DVD


You can get away with just mastering the same mix again, most likely. This would be especially true if you employ EBU R128 [0] and other such modern audio engineering practices. I've personally employed automated R128 dynamics matching for the past ten years, in order to output from one stereo PCM source file, a separate dynamically-matched master for each digital distributor's particular LUFS standard. i.e., YouTube's -13 dB LUFS, Tidal & Spotify's -14 dB LUFS, and iTunes' -16dB LUFS.

[0] https://www.sweetwater.com/insync/what-is-lufs-and-why-shoul...


It'd be cool if there was something akin to Dolby Atmos where you don't deliver a master, but rather individual tracks and associated metadata, such that it can then be mastered on the fly depending on the output equipment.

Many games let you specify the audio output, and then adjusts the mixing accordingly. This would be something similar I guess, but the mixing is probably best done at the streaming provider, to save bandwidth and other resources. (For static content like music it only really has to be done once per output type.)

I'd like something like this, I have a vastly different sound setup at home than on the go, but listen to mostly the same tracks. Most sound pretty good on my home setup, but on the go I often find myself adjusting volume up and down between tracks, despite things like normalization being on.


I think this is one of the things driving the vinyl resurgence. Albums that release on vinyl tend to less alteration to the dynamic range... Ymmv I've not listened to everything out there :-]


The real reason for that is because of the physical limitations of the medium. You have to be very careful when mastering for vinyl that you aren't knocking the stylus out of the groove by making things too loud (especially sudden changes are dangerous). Even for less proactively mastered stuff, if a master is received too loud, the factory will reduce the volume before pressing so that it is playable.


This could still be solved. Streaming services and digital music stores like Spotify and Apple Music could simply allow labels/artists to supply multiple masterings of the same tracks. One would be the default offering; the other would be a 24 bit file mastered with all loudness-optimising compression disabled. Popular music also routinely uses track level compression for artistic reasons and to balance the mix; it would be up to the artist whether or not to pare back any of that. Obviously the artist’s intent should take precedence here.

Then, as an end user, we could choose which we preferred. By default there would be no change in behaviour. Where software updates are available, an the option could be provided to swap between versions at will. For older devices, the streaming service could let the end user choose the high dynamic range version as an account-level default.

Personally I’d want ready access to both versions. When listening to music is the singular activity in a quiet environment, full dynamic range is great. But as soon as I’m not solely focused on music, multitasking or in a noisy environment, I’d actually prefer the compressed version.


24-bit for playback is a waste of space and bandwidth, 16-bit audio has plenty of dynamic range for any kind of music enjoyed by human beings.

In the old days, the recommendation for digital audio was to master for -20 dBFS average, use no compression (or very little) and let any peaks fall where they may in the 20 dB headroom. And nobody complained about the noise floor at ~76 dB below the average level.

I wish everyone would go back to mastering to that spec, instead of slamming everything to 0 dBFS with loads of compression and often clipping on top. Obviously still allowing use of compression for artistic reasons.


24 bits gives you more headroom to work with, making mastering easier. It also quells any concerns about 16 bits being insufficient. As for "waste of space", the simple fact is that we don't have a shortage of space when you're talking about audio files. Hard drives are routinely over 1TB now. Remember, the person you responded to proposed offering the 24-bit version as an optional "audiophile" version of a track, not the default offering. It's not like everyone will want to stream the top-quality version for listening on earbuds.


You’re correct that the file size of audio stopped being relevant many years ago. In a world of Netflix 4K streaming, even the most inefficient audio formats don’t move the needle.

(If it did matter, which it doesn’t, it occurs to me that you could even create a 24 bit file that only contains 18 significant bits [dithered to 18 bits then padded to 24] in such a way that the space was reclaimed by ALAC or FLAC encoding.)


I agree with all of that; I only suggest 24 bits in order to shut up the people who can’t get over the idea of using 16 bits for wide dynamic range content. I just want the better mastering. I don’t care whether it takes extra placebo to make it happen.

The bandwidth consequence would be trivial bordering on nil given that only a small number of albums would ever receive the treatment, and only a small number of end users would choose the option.


What about adding metadata to the audio file?

* https://en.wikipedia.org/wiki/ReplayGain

(I have no knowledge in this field, so am asking out of genuine curiosity.)


If you take a dynamic piece of music and aggressively compress it will create a weird pulse like effect. If you add compression/limiting to each individual track tastefully you can increase the overall volume without it being noticable.


In general, I kind of agree.

It's not possible to separately compress constituent stems in a mix. Once they're mixed, the impact that you can have on "loudness" is restricted to the whole, frequency-separated bands (multi-band compressor), or extracted parts. From an information-theory perspective, the available leeway for control over presentation is restricted once the audio is mixed.

However, I'd rather have a simple compressor running on playback than have a whole industry ruin their content.


That's way worse than loud mastering. There is no reasonable general method of compressing the mixbus at playtime, it depends on the source material and musical considerations. Its not even possible to find the right compression level without using your ears.

To be fair, I believe mastering is one of those domains that may soon be made obsolete with machine learning. But we're not quite there yet, and the existing methods require a lot of DSP power.


We also fail to protect ourselves from non-music hearing damage. Vacuums, lawnmowers, leaf blowers, traffic, etc will all degrade your hearing over time. I always wear ear protection whenever the sound level is even slightly uncomfortable. I'd rather spend my hearing on music than on noise.


Music in cars sounds absolutely terrible and the only stuff that sounds any good is generic pop music.


I disagree with this completely. Many midrange to high-end car sound systems are the perfect listening environment for many types of music, and come standard with excellent sound systems.


I guess if you listen to the top 40, any sound system is perfect.


"Nobody listens to top 40 music - it's too popular." - Yogi Berra, probably


What do you listen to?


Not all cars. The optional B&O high-end systems found in higher-end Audi, Mercedes and BMW are quite amazing, especially if fed a FLAC, Apt-X HD or other quality signal. These systems are expensive options, ranging from $3500 to $7000.


The quality of the sound system in the car is possibly secondary to the amount/quality of sound deadening in the car. Road noise is a nasty thing unless there is a lot of well engineered sound deadening material and acoustic design.

Once that's taken care of, then the audio can be appreciated. When I last owned a car, it was an enormous Lexus. The audio experience was sublime just with the premium factory system. But with the audio off, the cocoon was still so much better than what I've experienced in "regular" cars.

Unless you're in a Rolls Royce or Maybach while driving, I seriously doubt you'll tell the difference between FLAC and MP3. There's just too much external interference to compete with the subtleties that a high end audio signal can offer compared to a more typical system.


Regardless of what car you’re in, you’re not going to be able to reliably tell the difference between a flac file and an mp3 with a reasonable bit rate (256+).


I disagree. I have a stock sound system that I can easily tell the difference between a 320kbps MP3 and a CD flac in a blind test. I don't believe I have any special hearing as pretty much anyone I'm with can tell the difference.


-It does, of course, also depend on which MP3 encoder and settings you use; I did a most interesting (to me, anyway!) series of ABX trials on my own ears a few years ago.

From a 24/96 PCM master I prepared lots of transcodings - 16/44.1, 12/44.1, 8/44.1, 16/22, 12/22, 8/22, 320, 256, 192, 160, 128kbps mp3 with various options for encoding, ditto AAC - etc, etc.

I then used the (brilliant!) ABX plugin for Foobar and an excellent set of headphones, and after a few days I had a fair idea of how good my ears really were. Or, as it turns out, weren't. In most cases, 256kbps mp3 and red book were impossible to tell apart; 320kbps being indistinguishable in all but one case, if memory serves. However, at both 256 and 320kbps I had to listen hard for encoding artifacts rather than just enjoying the music.

Most tracks were eminently listenable at 192kbps mp3, but I could then reliably tell whether I listened to an mp3 or PCM without much effort.

Hard disk cost being what it is, I still ripped all my CDs lossless, though.

128kbps? I shudder at the thought of my 20-year-old-self telling myself and others that it was just about as good as uncompressed audio...

Oh, and you'd be surprised at how little dynamic range is actually utilized; on most tracks, telling 16/44.1 from 8/44.1 was next to impossible.


Yeah I've messed with dynamic range a bit, even got caught by the 24/96 hype for a while. But after running my own blind tests, I couldn't tell any difference with it.


On some songs with some setups I do believe you can still hear a difference but it's unlikely you'll have those songs and/or those setups. The vast majority of the time, unless the encoder for the MP3 is garbage, you won't be able to tell them apart.

https://www.head-fi.org/threads/abx-test-of-320kbps-vs-flac-...

https://blog.codinghorror.com/concluding-the-great-mp3-bitra...


Lower-fidelity playback systems and environments can actually reveal artifacts in lossily compressed audio. The psychoacoustic encoder makes assumptions about what noises mask other noises - if noises it expects to be present are absent (for instance, if your playback system has a poor frequency response), then low-bitrate noises can be unmasked.

I would not be surprised if additional noise, such as road noise, can also subvert masking phenomena in some way, but I don't know of any studies that tested this.


Its not just the speakers, its the loud rumble from the car moving. Maybe some cars have perfect sound insulation but I would say thats not very safe to be driving if you can't hear the outside.


nitpick: apt-x hd is a lossy audio codec and is almost never the source encoding of the music. most people are streaming aac, vorbis, or mp3, which is transcoded on the fly to apt-x hd. this is a lossy-to-lossy transcode which is basically never going to be transparent. apt-x hd is not a "quality signal" unless you happen to have a bunch of flacs on your phone.


I meant apt-x HD as the transmission codec for a Bluetooth connection rather than as a compression format for the audio files themselves.


I understand. what I'm saying is, nine times out of ten, the audio files themselves are already going to be in a lossy format. it's not really practical to have lossless content on your phone or stream it over a mobile connection. the lossy-to-lossy transcode is going to sound bad enough that most people would notice in a double blind test. this isn't some mp3 v0 vs flac audiophile nonsense; lossy-to-lossy transcodes sound noticeably bad.

from your other comment, I see that you are using a pretty good DAC in your car. kudos to you, sir.


>especially if fed a FLAC, Apt-X HD or other quality signal.

How many cars actually support FLAC or Apt-X HD?


All those with an analog or USB input?


No.


It's how I do it. I use a Sony NW-WM1A and plug it into the analog input in the car. Sounds fantastic.


You seem to have basically no idea of what you're talking about.


Really? I seem to not know what sounds good to me? Okay, then.


Classical sounded great in my 2011 Honda Fit Sport. Maybe I have low standards though.


Hey, that's what I drive as well. It's surely not "audiophile quality" but it's got just enough range that most music I listen to sounds surprisingly good for a stock sound system (as long as the source file is high quality).


This is a great point. When I'm listening on laptop speakers, every little bit of extra loudness helps.


I used an app called Boom for years to boost the volume, when I only had my laptop around and want to boost the volume. I haven't used it in a while however as my recent Macbook has much better speakers so I'm not sure if the v2 update is as good as v1.

https://apps.apple.com/us/app/boom2-volume-boost-equalizer/i...


Ian Shepherd -- mentioned in the article as the founder of "Dynamic Range Day" -- is one of the great warriors in this battle. He's an extremely knowledgeable mastering engineer (among the darker arts in audio), and has really done a lot to help turn the tide on massively over-compressed audio.

But I came here, actually, to put in a plug for a plug(in). He worked with some folks to create a VST called "Loundness Penalty." You throw that on your master channel, and it will tell you what the various streaming services will do to your audio (for example, by how many dB Spotify, Apple Music, You Tube, and so forth will turn your audio up or down based on the integrated LUFS reading).

Even more impressive is a plugin he worked on called "Perception" that can avoid the inevitable psychoacoustic bias (to which we are all naturally subject) of thinking that because this track with this effect chain is louder than without it (or with another effect chain) it must be "better."

These are commercial products, available from an outfit called MeterPlugs (https://www.meterplugs.com/).

I should say that I've never met Ian, and have no connection to this company. I do work with audio in a professional context, however.

I should also say that I really think the tide is finally turning on all of this. I don't know that the "loudness wars" are absolutely in the rear-view mirror, but I think there's a lot more awareness of the issue. And I also think that fly-by-night mastering engineers who proceed to crush the hell out of your tracks with limiters and whatnot don't get as many repeat customers as they once did. And there's better metering in general for getting a track to have proper dynamics without having the listener feel the constant need to turn the volume up and down.


I run a recording label that publishes Drum&Bass music (RuffAndTuffRecordings.Com), (Drum & Bass music currently is known as one of the loudest engineered music genres put out on the block right now). I went the other way in terms of our releases... Our music is only as loud as it needs to be. If a listener wants it louder, they can simply turn up the volume.

Going to nights where these super-compressed tracks are often played at unimaginable volume levels, You must wear ear plugs, otherwise you're guaranteed to go deaf.

When music is compressed, dynamic range is lost, and the actual elements in each track like bass and ambient sounds are limited in how they can create a mood in a track. These loud tracks are impressive in terms of grabbing attention, but over time they create fatigue for listeners, and people tend to not want to listen to additional music after a few loud tracks.

There is a high price to pay for continuing down the path of loudness with music.

The main driver for ending the race in music should be doing whats right, music should have dynamic range as the main focal points of audio engineering. Until that gets sorted out, bring your ear plugs.


I think your point about "listener fatigue" is important, because that's a big part of this. It's not just about the way a squashed track won't sound as good or as interesting (though it won't, of course).

And yeah, that's a famously "loud" genre. What drives me insane is going to, say, a mostly acoustic live show and having the person behind the desk mix it as if it's Drum & Bass. I never ever go to a show without earplugs, because you just never know.


> it will tell you what the various streaming services will do to your audio (for example, by how many dB Spotify, Apple Music, You Tube, and so forth will turn your audio up or down based on the integrated LUFS reading).

is there a summary anywhere of which services are better and worse in this department?


Check out the manual for Loudness Penalty. It has a table that says what the different services do.

So, for example, Spotify enables normalization by default, they target -14 LUFS, use limiters, normalize both tracks and albums, and will turn up quiet tracks. Amazon, by contrast, does not use limiters, does not turn quiet tracks up, only normalizes at the track level, and so forth . . .


Daft Punk's Random Access Memories is mentioned in the article, and it really does have good dynamic range .. for the 2010's

But compared to an album like Spies 1988 Music of Espionage which has a dynamic range that used the most of the CD format (especially track 3 - "Interlude" -- listen for the drum hits at 0:39 after the organ chords), it just isn't (technically) comparable.

I have stopped buying older CDs off Amazon, switching to Discogs so that I can be sure to get an original pressing - made before it had been "remastered".


I wonder if more dynamic music is actually worse for developing tinnitus and so on when listening on headphones. As the perceived loudness is lower for dynamic music, people turn it up to compensate. This means that the peaks in real terms are actually louder than the equivalent super-compressed stuff while achieving the same listening level.


I dunno - I'd want to consult an audiologist or ENT doctor (otolaryngologist) before answering. But my uneducated opinion is that dynamic music gives the ear time to recover from the loud sections (given normal listening levels), while the compressed-to-all-hell music does not, since it's a constant high level sound.

My tinnitus came from being around jets & equipment in the military, concerts (Rush's Presto tour in 1990 was super loud), and competitive shooting. I double-up on my hearing protection these days.


My spouse's father is a well known (Grammy nominated) mastering engineer.

He hated the loudness wars. It was, often times, bands and labels insisting that the albums be "louder" than whatever other comparable album they were trying to emulate, and thus it just escalated with no real sense until it just got ridiculous.


Eventually this gets us to here: https://hitchhikers.fandom.com/wiki/Disaster_Area



"ReplayGain is a proposed standard published by David Robinson in 2001 to measure the perceived loudness of audio in computer audio formats such as MP3 and Ogg Vorbis. It allows media players to normalize loudness for individual tracks or albums. This avoids the common problem of having to manually adjust volume levels between tracks when playing audio files from albums that have been mastered at different loudness levels."

https://en.wikipedia.org/wiki/ReplayGain


This doesn’t solve the problem that “louder” tracks typically have substantially reduced dynamic range Such tracks are often irreversibly worse than those with less compressed range.


Of course not, it's a stick. If widely adopted, it eliminates the incentive for the extra compression in the first place.


Watching without headphones feels impossible nowadays. The talking volume is low, but the music is top whack. So you need to keep turning the volume down each time there is music otherwise it is embarrassingly/annoyingly loud.


I think you're talking about TV/Movies, but the article is talking about music.

Most home audio receivers have user-configurable dynamic range compression, which compensates for the "quiet voice, loud music" problem. You can often also boost only the center channel, which helps.


This is a consistent bugbear of mine, and I think a big factor in the resurgence of vinyl. I find it frustrating I can take a recording from a format with limited dynamic range and stereo separation, which is incredibly vulnerable to environmental contamination and vibration, where the playback device adds its own bundle of mechanical noise and unusual non-linear responses... and yet if I have a good pressing it will sound better than a "digital remaster" CD which has none of these inherent issues, because the digital copy has been compressed and limited to the point where it's fatiguing to listen to.

That said, I've bought a few modern albums which have been really nicely mastered; they've got that rich, deep '70s LP sound but available on a FLAC download without all the surface noise and rumble.

Also that section on the '60s loudness war underplays quite how LOUD some of those old mono 45s are... I typically set things up so my peaks are at -12dB when recording, and I have the odd mid-1960s single which will be pegged right up against the red if I don't adjust the levels. No wonder there were some pressings that were notorious for throwing the stylus out of the groove when people started chasing ultra-low tracking weights in the early '70s.


Depends how recently the vinyl was pressed :P

Dnb vinyl was very popular for a long time (still has its fans), and it was very much part of the loudness war. There are a couple of producers who were part of my record collection that I sometimes actually would not play because I couldn't be bothered dealing with the ridiculous over-mastered loudness in the track compared to what I was trying to mix them into.


"Rage Against The Machine" exemplifies the victory in this war: they made clipping integral to the music. I wonder if they're still waiting for the world to get the joke.


Clipping individual parts doesn't sound the same as clipping in mastering. If you clip a single harmonic tone (in the musical sense, i.e. a fundamental + partials at integer multiples of its frequency) you only get harmonic distortion. All new partials generated are still integer multiple of the fundamental. The timbre changes but it doesn't sound dissonant.

When you clip a more complex waveform, you can generate new partials with non-integer multiples of the fundamental frequencies, which sound dissonant. If it's simple enough (e.g. a guitar chord), the dissonance will be small enough that it still sounds musical, but if you clip something complex (e.g. an entire mixed-down track), it will just sound like noise.

See https://en.wikipedia.org/wiki/Intermodulation

Clipping of individual parts is integral to RATM's music, but I don't think clipping of the whole thing at once is. See their entries in the Dynamic Range Database:

http://dr.loudness-war.info/album/list?artist=rage+against+t...

I haven't actually heard the SACD release of Evil Empire, but with those numbers I'd be surprised if it was noticeably clipped in mastering. (Note that vinyl mastering can increase the DR measurement without audible changes in sound, see: http://wiki.hydrogenaud.io/index.php?title=Myths_(Vinyl)#Eff... , so I'm much less confident about the vinyl releases with similar numbers.)


Pretty funny to listen to one of the lowest-rated albums in that database in terms of dynamic range, Gray Data by Five Star Hotel: https://open.spotify.com/album/6Q8hkb14PqvyLVEdmWXiD9?si=H5w...

Makes RATM sound like easy listening. Which I suppose it is these days anyway.


...my go-to-album when the perils of excessive dynamic range is to be avoided (cough) is Neil Young's 'Weld' live album.

It basically has two states: 'on' and 'off'.

However, it DOES serve an artistic purpose.

If you get a decent pressing of Motörhead's 'No Sleep 'til Hammersmith', though, you'd be surprised at the dynamic range actually offered, contrary to popular (mis)conceptions about the Motörhead sound... :)


First page of that database Bassnectar :-)



Sorting I noticed there were 2 that got "00 00 00" ratings. One basically sounded like white noise and the other lived up to my hopes and had me laughing https://www.youtube.com/watch?v=prXY-ZQdKJg

Unfortunate the site doesn't have a graph of loudness vs time.


That's actually amazing music, though will blow your speakers at 1% volume ;)


Somewhere on the shelf I have the Erich Kunzel/Cincinnati Pops performance of the 1812 overture.

They used real cannons, and the recording & mastering engineers did their best to reproduce them faithfully. It is one of very few albums I have with a prominent warning sticker on the album which is there for a good reason, not for bragging rights - it basically says that unless you are quite conservative with the volume knob, your speakers will die once the cannons do their thing.


Interesting. The cannons would be so loud that, if they're at 0db, the rest of the orchestra would be waaay down there without substantial dynamic range compression. You'd have to crank the amplifier waaay up to listen to the orchestra at seemingly normal volume...


-Obviously there must be some DR compression, however the recording goes to great lengths to ensure Tchaikovsky's intent is preserved - that the cannons drown out the orchestra and is more or less a physical experience.

Your comment has prompted me to do something I've been thinking of doing since I got that album - open it in Aucacity and see what the DR actually is - during the 1812 crescendo, it peaks at -0.02dB[FS]; RMS is -24dB.


It's an interesting thing. When you participate in this war as a creator of recorded music, as I do, you're sacrificing dynamic range so that your track fits in with everyone else's. For some types of music, it's a trade-off worth making, for others, not so much. Getting close to what the pros do is pretty straightforward if you have experience and a real mastering tool. Those tools are getting smart about analyzing your track and building an eq/compressor/limiter/maximizer stack that rocks, but the gods of top-40, for example, can make real magic happen when it comes to loud-but-not-distorted.


It's not a trade-off worth making at all from my perspective. I have a volume knob, pretty much all audio software has an option for volume normalization, radio DJs can apply something similar if they wish.


I say this with all due respect, but your copy of Ozone is getting you nowhere near what the pros do. Not even the same planet.


All he said is that the "tools are getting smart[er]," and he's right.

I still can't bring myself to hit "Mastering Assistant" (still less, "Mix Assistant" in Neutron) and call it a day. But honestly, it's pretty spooky how good these tools are getting. I'm particularly impressed by the way it figures out where to set dynamic EQ points on problematic frequencies. I haven't played with Gullfoss yet, but the pros I talk to say it's even spookier.

Of course, what you may mean is that there's no substitute for a $50,000 Neve console, a treated, isolated room, monitors that cost considerably more than your car, and an engineer with twenty years of experience. Sure.

But honestly, it is reasonable to wonder whether this is a bit like when we used to say, "Well, a computer will never beat a human at chess."

It's also not true that no professional mastering engineers use Ozone. Most big shops at least have it lying around (if only for that limiter, which many pros emphatically do use). It's also sometimes the right tool for the job. And there are a few serious pros for whom it's their main tool. Not many, granted. But interface aside, it's not clear that when it comes to "clean" processing, that FabFilter's stuff (which is very widely used) is any better.


Thanks for mentioning those other plugins. I'm not serious about mastering (it's about the tune!), but now I'm curious, and the FabFilter mastering bundle may have to go on the wish list.


Well, perceived loudness is a tough nut to crack for sure, but that's what's in question, and my evidence is here: https://soundcloud.com/willhaslett Enjoy! Same "league", "planet", whatever. My work is plenty loud when I want it to be :)


The loudness war (or loudness race) refers to the trend of increasing audio levels in recorded music, which reduces audio fidelity, and according to many critics, listener enjoyment.


WHY does this happen? Does loud music sell better than silent music? Nobody here, not even Wikipedia, mentions the reason WHY it happens.


From the History section of the Wikipedia page:

Jukeboxes became popular in the 1940s and were often set to a predetermined level by the owner, so any record that was mastered louder than the others would stand out. Similarly, starting in the 1950s, producers would request louder 7-inch singles so that songs would stand out when auditioned by program directors for radio stations.[1] In particular, many Motown records pushed the limits of how loud records could be made; according to one of their engineers, they were "notorious for cutting some of the hottest 45s in the industry."[3] In the 1960s and 1970s, compilation albums of hits by multiple different artists became popular, and if artists and producers found their song was quieter than others on the compilation, they would insist that their song be remastered to be competitive.


And if we go further back we find that there was a sort of brightness war in orchestra tuning going on a long time ago. Think they standardised on 440 Hz by now?


I remember watching a video by a music theorist on YouTube (12Tone) where he mentioned that laws were made to define what Concert A would be - https://www.youtube.com/watch?v=BzznBt8tVnI


Noisia (dnb producers) are notorious for this...

I had a couple of Noisia records that sounded ridiculously louder than all my other music. I guess it's part of the "authentic" vinyl mixing experience to have to manually adjust a turntable trim mid-mix to avoid it sounding ridiculous.


You can say a lot about Noisia, but their mixes always sound perfectly clean (if you can even say that about neurofunk). Not like Metallica's Death Magnetic album for example, for which the compression artifacts really hamper the listening experience.


True. Also their electro was also very colourful and vibrant sounding stuff. It's just annoying when you're in the middle of a mix, you don't check how the total mix sounds in your headphones, bring in the Noisia track and the other track is almost drowned out :D Of course a good DJ knows his music and will be ready to make adjustments, but it would be cool if the loudness war wasn't a thing


A very detailed article about dynamic range and the Loudness War: https://www.soundonsound.com/sound-advice/dynamic-range-loud...


One thing the article doesn't mention is mastering for listening environment context: Classical music that takes full advantage of dynamic range is very easy to listen to in a quiet room; but difficult to listen to in a car or airplane. This is because the quiet (delicate) parts get drowned out by road / air noise.

Anyway, this is one of the concepts that is forgotten when people discuss music in 5.1: Part of the point is that a 5.1 mix can be a "living room" mix, where the engineer can make an uncompromised mix, and leave the compromises for listening in a car, or with crappy headphones, in the stereo mix.


On Netflix, I found a trick to have the full dynamic range of sound: You have to keep the volume to a minimum in the netflix player, and to compensate, have the volume at maximum in your computer. I am very sensitive to the respect of the dynamic range, this is very important in order to have pleasure listening. It is the same problem with photos and video games : In an "HDR" treated image, the sky is much to dark, leading to a non realistic image.


Wait, are you saying that Netflix alters the dynamic range of its source content? That would be awful unless used as a clearly labeled option.

I always felt that Netflix' audio quality was rather poor and lacked punch as well as dynamic range in a proper home cinema setup, but assumed it was due to their codec or streaming technology choice. However now that you mention it, it might as well be deliberate to optimize for subpar watching environments (laptop speakers, soundbars, etc.) - indeed it sounds as if their content was mastered for TV rather than cinema.


I always loved this piece on this topic: http://ozzgod.com/dynamicrange/death.html, which has been moving around the internet (my old link was dead, and even it was a copy of the original) for as long as I can remember.


Reminds me of compiling ACCGain to normalize tracks from Apple store in Amarok: https://jonasfj.dk/2007/08/volume-normalization-with-amarok/

These days I just use Spotify :)


Reminds me of "DJ Double R The Loudness King" aka Rick Rubin, probably one of the most famous producers these days/the last 20-30 years: https://en.wikipedia.org/wiki/Rick_Rubin


I think there's another type of 'loudness war' going on with how we communicate. One of my peeves in particular the catastrophication of minor insults as 'hate', as I feel it's diluting our concern over the actual atrocities that happen on the other end of the spectrum.


At burning man and etc the last few years it's really seemed like some of the music makers are just trying to see what the speakers can do.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: