Music streaming services have since become a very important music listening medium. Spotify, Google Music, Apple Music and others do normalization of these loudness levels. This neutralizes the loudness wars as it makes loudness wars treatment of music useless, at least when listened on streaming services [1]. Music mastered in the loudness war will still have problems with dynamic range but the perverse motivations causing the war are simply not as relevant for new music.

I agree the loudness war was a huge problem a few years ago, but changing trends in the ways music is being listened to is solving it. The way you present the loudness war problem is therefore somewhat out of touch.

[1]: https://motherboard.vice.com/en_us/article/ywgeek/why-spotif...

It is especially noticeable if one listens to "modern" vs older CDs vs the radio. I suppose Producers now feel compelled to exercise subwoofers and thump the audience, or compensate for crappy earbuds in the target audience, or maybe that's just what people want because even radio commercials are complicit -- it's difficult to listen to someone talk with "thump/thump/thump" drumbeats in the background. Often, I simply turn it off.

In my car, I have the bass response dialed back by about 50%, and even dip the mid-range by about 15% to get what sounds to my ears like a flat response.

I have mild hearing loss from the 250 to 1k range, so my ears are already attenuating the signal -- I can only imagine how "bass heavy" it must sound to someone with normal hearing!

There is nothing inherently good or bad in music production, no timeless rules as to how much treble or bass, compression, distortion, reverb or anything else you need. It has all been done to the extreme and what is deemed good is subject to constant change.

My personal example of a song where it doesn't work is Johnny Cash's "Hurt". Around 3 minutes into the song, Johnny Cash's vocal noticeably distorts. From my perspective, the distortion is absolutely a loudness war phenomenon: if you look at the waveform in an editor, the song starts off pretty "hot" given that there's a very noticeable crescendos at the end. At the end of the song, where the music is pretty much maxing out, there is no bandwidth left. There is no other option to stand out but to push Johnny's voice to distortion.

I've seen people in message boards say the distortion in the vocals adds "intensity". Personally, I'd love to see an un-Loudness War version where Johnny Cash gets several dB more to cleanly sing over the music. Where dynamics, not distortion, is what is driving the intensity. For my tastes, it would be much preferred.

[1]: https://en.wikipedia.org/wiki/Loudness_war#Examples_of_.22lo... (not including the 'remasters')

I found the album a bit tiring to listen to because of the continuous loudness. No particular parts really stood out to me, because everything was just sort of loud and it just seemed to go in one ear, out the other. I could pay real close attention, one can always listen better :) But again it is a bit tiring and you can't do much else.

Then again, I just really prefer their first 2-3 albums, which have quite a different sound altogether.

And I'm curious which Oizo albums you're referring to? I love his stuff, and yes some of his tracks are quite loud, (but not all the time, the whole track), but they never quite struck me as a typical "loudness war" type of loudness. Unless I'm thinking of the wrong tracks here (No Day Massacre? Last Night a DJ Killed My Dog?), he seems to like to hit a well sound-designed overdriven bass buzz, with not too much else playing at the same time (and where it does, close attention to envelopes, cutting off hits, ms transient timings) if you do that right and just normalize to max amplitude, you're 95% of the way there (at which point my experience is that compression on top of that usually fucks up that careful detail work, but maybe I need more practice or a different plugin). Possibly I'm thinking of the wrong tracks here, at least you gave me an reason to re-listen his stuff with a different ear/attention (loudness sound-design), which is always interesting :)

You'll need an exceptional good clock to start with, and all other equipment needs to align to that clock. Then all plugins/processing you use needs to be in the same 24/192 domain, otherwise your signal is reduced to the limit of that plugin/processing and all previous efforts are lost.

Most music producers use samples, most are 16/44, so what's the point to try to get that to 24/192, filling the signal with zero's..

If a piece of music is in very rare occasion truly 24/192 then the listener who downloaded the track still needs a exceptional good clock (that are both expensive and hard to find) to playback without signal reduction.

IMAO 24/192 is just a marketing thing for audiophiles that don't really understand the implications. 24/96 should be a reasonable limit for now, although personally I think 24/48 is enough for very high quality audio.

> Most music producers use samples...

Most people interested in better-quality sound in this particular context aren't listening to contemporary electronic music with samples. 24/192 or SACD is so desirable for reissues of older recordings in pop or jazz genres where those formats were mastered with higher dynamic range, while the available CD versions or lower-bitrate downloads were mastered with loudness-wars compression. The format is also attractive to classical music listeners, because SACD gives you multichannel audio; and some classical labels are now giving loudness-wars treatment to the non-SACD or non-24/192 formats of a particular new release.

That's depressing.

In the studio, I would say that 24 bit at least should be the norm for recording purposes.

24 bit recording gives you very noticeable increased headroom (about 20dB). This gives you quite a bit more flexibility recording lower levels without concerning yourself about the noise floor. The difference isn't huge for most prosumer setups in practice, but given that the processing power and storage power of computers makes recording in 24 bit trivial to do, there really is no reason not to record 24 bit these days IMHO.

Sample rate also comes into play, mainly if you have older plugins that do not oversample. Some of the mathematical calculations involved, particularly if they are quickly responding to audio changes (eg limiting / compression, distortion), or are using "naive" aliasing-prone algorithms (eg naive sawtooth wave vs. something like BLEP / PolyBLEP etc.), can introduce frequencies beyond the Nyquist that may translate into aliasing. These days, I would say most plugins do oversample internally or at the very least give you the option to do so. There's also a VST wrapper to over-sample older plugins as well (http://www.experimentalscene.com/software/antialias/). So I do not think recording over 44.1kHz is very necessary these days. I don't discount opinions from people that recording at 192kHz "sounds better", though, given the possibility that they are using plugins that are prone to aliasing at 44.1kHz rates.

I personally do not see any benefit of 16/44.1kHz for playback most recordings. Maybe 24 bit would be useful for heavily dynamic music (one of the few categories where you generally find this is orchestral music), but I'm thinking even for here the 96dB range of 16 bit audio should be enough for most cases.

To be fair, that only applies to the digital part of your signal chain. The analog portion is going to have nowhere near 24 bits of room above the noise floor.

The article is pretty clear that 24/192 can be reasonable for production-- it's just not reasonable for playback.

But your arguments aren't quite right, IMO. If you have a 16/44 sample, and you don't play it at full volume, you get some use out of those extra bits. Especially if you have a volume envelope.

Also many modern samples are actually saved as 24 (or 32 bit even). Especially if they're my own creation from noodling around with softsynths, but they're shared like that as well, obviously.

Then, if you apply a plugin or effect that supports 192/24 output, on a 16/44 sample, you still get useful information in those additional bits, even if the sample did not. Think of the very quiet end of a reverb tail, for instance.

But that's for producers. It's always good to have your working stuff in high quality, you never know how far you'll end up amplifying those low order bits.

So I can see the use for 24 bit audio (in certain contexts), but I'm really not so sure at all what the 192kHz is good for. Since it's waaaayyy above the human hearing range, all I can think of is dithering. You can hide a lot of your dithering noise in the ultrasonic range (which almost seems like free lunch IMHO) and then ... you obtained even more (virtual) bits of fidelity! Which you didn't really need cause you were using 24 bit audio in the first place.

I agree it's mostly a marketing gimmick, otherwise.

    the 24/192 downloads or SACD releases are often the only way to hear the album with real dynamic range.

Also true? The completely absurd fact that sometimes a vinyl rip of an album will actually have the highest dynamic range, even though vinyl has a much smaller dynamic range than 16/44 audio. Bands often use "loudness wars" mastering for the digital release and then proper mastering for the vinyl release.

Back in the day you'd occasionally see remastered "gold" discs released. The advertising made a big deal about the disc material. Those probably sounded different too (they managed to sell them at a great premium) but with those, they sounded better because of the newer remastering, not the disc technology.

It's certainly possible this is the case with some of those releases remastered for SACD as well. The label probably didn't give Green Day a huge amount of money for production when they made Dookie, for example, but it eventually sold 20 million copies and additional production made sense. If it sounds better to the listener it's a real benefit, too, but it is quite likely not down to the playback technology.

Two oddball things I've noticed about remasters in the last couple of years: there's some agreement out there that the newest remaster of Hendrix's records is not the best. And King Crimson, who has an audiophile following and decorates their catalog with remasters with alarming regularity, removed a little dynamic range (an oversimplification) from the newest remaster of Larks Tongues in Aspic when remastering the CD and mastering for DVD/A because people very understandably complained about the (technically very good) previous version being too quiet. Audiophiles say they want dynamic range, but...

    The label probably didn't give Green Day a huge amount of
    money for production when they made Dookie, for example,
    but it eventually sold 20 million copies and additional 
    production made sense.

No. It's still entirely silly. The reason? The dynamics are unrecoverable once squashed by a limiter or compressor during the remastering process. The fidelity of the delivery medium is moot after that happens.

If the SA cd promised that they reloaded the source tape/protools/whatever DAW was used and remixed/ remastered the songs to actually have dynamic range then I would be interested. As far as I am aware this isn't happening, and is implausible for any record of considered a classic

Many SACD reissues do go back to the source tape. This is a frequent cause of complaint with SACD reissues of classic jazz recordings from the 1960s: sometimes you get better sound in terms of dynamic range than any previous CD issue of that recording, but in the meantime the source tape may have deteriorated in parts.

Even with recordings from the “loudness wars”, there is sometimes room for dynamic range improvement when remastering. A good example is Rush’s album Vapor Trails. This was an infamously botched recording upon its original release, on a scale with Death Magnetic. Because loudness-wars treatment plagued the original tracks before mixing, the damage could never entirely be repaired. However, the additional process of compression applied to the source during transfer to CD could be reversed, so the album was eventually reissued as Vapor Trails Remixed, and while still flawed, that reissue has a lot more room to breathe than the original CD release.

"Remastered" reissues on the other formats – CD or lower-bitrate downloads – are nowadays expected to be listened to through cheap earbuds or speakers and perhaps in noisy urban environments. So, the engineer applies "loudness wars" treatment, compressing their dynamic range so the listener can still hear the quiet parts even with all the noise around them.

The other poster is not saying otherwise, though. They're just saying that "hi-def" formats, while technically unnecessary for end-user listening, are often the only way to obtain a decently-mastered recording.

There's no technical reason for things to be that way. But that's how things are.

It's sort of like buying a new car. You want the more powerful engine? Well then you also have to buy the package with heated seats, the moonroof, and the fog lights. There's no technical reason for it, but that's the only way they'll sell it to you.

It's also what this entire discussion's about.

Of course formats above 44.1/16 are useful for professional work; nobody's ever said otherwise. Just like graphics professionals work with high-res images and lossless formats even if the end product is a .jpg on a web page.

I noticed a similar thing with TV a few years ago. Despite watching a standard def. channel on a SD TV some programmes had a very noticeably better image quality. I think these had been shot and edited in HD and although the 'last mile' was still in SD there had been less degradation in the processing so the final picture was much better.

the mastering engineer has to be approved and there are some minimum dynamic range standards

also 24 bit (but not 192khz) master files have to be supplied

reportedly some of the streaming services (Spotify, YouTube) are now applying some 'loudness normalisation' which will bring some of the bad loud masters into line (it won't restore their dynamic range but will make them similar loudness to other tracks)

the loudness wars were never about what's good for listeners, but rather a competition for tracks to sound louder than other people's tracks when played consecutively on radio or in your music player

and the iTunes files are 256 kbps AACs, you can't hear the compression

remember that 'compression' in this context is data compression and not audio compression (which acts directly on the dynamic range of the source)

I can most certainly hear the compression when compared with CD digital audio.

I fully understand the difference between data compression and dynamic range (not audio) compression.

What I'm saying is, lossy data compressed audio formats are already compromised enough to rule it out as a medium for audiophile use. Worrying about the dynamic range at that point is moot. It's going to be played on tiny, terrible sounding speakers.

with 256kbps AAC, really? yeah MP3 is old and even at 320kbps it throws away everything above 18kHz (which I can't hear personally, but some people can). However AAC is newer and better and blows MP3 out of the water (so do OGG and OPUS, btw). We got a lot better at psychoacoustics since MP3 started out. I strongly doubt you could hear it in a proper A/B test.

Then, 24/192 is mostly a "weak signal" to help you estimate if the audio was treated with care.

16-bits = 2^16 * 0.01 = 655 discrete levels

24-bits = 2^24 * 0.01 = 168000 discrete levels

To help you understand, imagine 1-bit audio. What would it sound like? For each frequency, you can only have a single volume, i.e. it's maximally compressed.

I'm not sure the 1-bit extreme helps with understanding this, particularly because the audio we'd be dealing with will be a mix of many frequencies; the waveform is an aggregate of the overall 'signal' at each frequency present in the mix; this is one of the reasons why multiband rather than single-band compression has become so popular (as it allows you to get the 'max' in each frequency band rather than one band dominating the overall result of the compression).

I think there's a difference to take into account when considering any given momentary sample and the overall effect - yes, compression does reduce the dynamic range of the music, but you would need some sort of variable bit depth which was linked to the instantaneous amount of compression being applied to get any kind of workable lower-bit-depth encoding scheme, which seems like a lot of complexity for no significant gain (to me?).

> Compression means it's effectively using less of the [dynamic] range and the dynamics [range] could be represented with fewer bits in an optimal encoding.

Lower DNR can indeed be encoded with fewer bits per sample.

The meme "Math class is hard, let's go shopping" is only slightly apocryphal. Two of the voicebox lines were "Math class is tough" and "Want to go shopping?"

Because we only have 8 bits per channel, we all can see banding in (dynamically) compressed images. That's what increasing the depth improves.

You cannot buy good recordings in other formats, in this format you can. So there's a market - not a very big one as such, and maybe created for all the wrong reasons, but it is there.

The problem here is that the "you" in this sentence is not "me", or even an entity I can really influence, much less control.

It's not really even the audio engineer whose boss is telling him to turn up the volume so high that only the top few bits of the 16-bit recording are in use. The boss is saying this so that the song can be heard when the user turns the volume down because of the obnoxiously noisy commercials on the radio. Those commercials have found that they're more effective when they turn the volume way, way up. And they don't give a whit about the quality of their audio as long as they can sell more cars or tickets or whatever, much less the quality of the songs that play after them, much much less the quality of the songs that someone downloads to listen to offline in a nice listening environment without commercials.

The solution isn't just "turn down your compressor ratio", there's a big, hairy Nash equilibrium/politics problem that can be bypassed by offering 24/192 as a secondary product. If you want to remaster it to 16/48 after downloading, you're welcome to do so.

Doing this process in 24 bit gives you a large margin of error to play with. No real point to keeping that for the recording people are going to listen to

Yes, those high dynamic range releases usually get published in 24/192. No, the fact that they get published in 24/192 does not, as far as human hearing goes, add anything to the dynamic range or otherwise the fidelity of the recording.

Since the correlation is so strong, it is of course entirely understandable that people assume causation exists.

Is there a technical reason why the mastering is so different for the two mediums, CD versus vinyl?

More often than not these days, the same compressed master is used for the vinyl. To combat the groove-jumping problem, the overall level is simply dropped.

Thus, people started mastering CDs for loudness.

An alternative idea of mine is simpler. Loud music is considered worse than quiet music (quiet music sounds worse, but loud music still does, and also bothers other people). So, when you need to pick a volume setting for your collection, you bias towards setting it lower, so the really loud ones don't become too loud. Thus, the quieter CDs are annoying because they always sound quiet, whilst the load CDs sound about right, because your volume is much more suitable for them.

I guess special purpose releases don't usually end up on the radio so they can be mastered for people who actually appreciate music. ;)

As ever this difference can be impossible to detect if the equipment and environment aren't of a sufficient fidelity/quality.

What kind of gaps? There are no gaps. Sampled signals perfectly represent the original wave up to half the sampling frequency. Analog systems are inevitably limited in their frequency response as well, so, given the same bandwidth, there would be no difference at all.

In the real world, imperfect A/D and D/A conversions are typically still far less destructive than all the mechanical and electromagnetic sources of noise that affect analog systems. You can't consider one but not the other.

https://www.youtube.com/watch?v=cIQ9IXSUzuM

I think you're right that recording equipment has a long way to go, though; regardless of format I think people can relatively easily distinguish real acoustical instruments from recordings.

Although in any case, nothing justifies the pricetag dished out to audio enthusiasts.

It's only at higher tiers that the difference might really be imperceptible to many ears.

A 200$ pickup is certainly better than a 20$ needle, maybe even 10x better The difference between a 200$ pickup and one that goes for 2000$ is miniscule. There certainly is a difference, but it's never as big as between the 20$ and the 200$ model.

That said, there are listeners who believe in the value of a 2000$ pickup and derive a lot of enjoyment from the difference to a lesser model. Who am I to say they're wrong?

Now when it comes to a manufacturer of very expensive cables (for example): Don't make me laugh...

Two things come to mind though: voltage/current delivery of your amp, and damping ratio.

The first depends on the characteristics of your amplifier. Some are better at delivering current, some voltage. The lower impedance (32) is better suited to high-current/low-voltage sources, which includes most portable devices, phones, etc. Conversely, the higher impedance (250) is better for high-voltage/low-current sources like tube amps.

The second is about the ratio of the headphone impedance to the amp output impedance. You want a high ratio, so if your source has a large output impedance then the higher impedance headphones will sound better. Good headphone amps sometimes specify the output impedance, or you can measure it.

Headphone outputs of mixers fall into this category as well. Proper audio interfaces and sound cards have no issues at all driving 250 Ω to deafening volumes. Laptops no issues as well (for me).