The Dark Art of Mastering Music | Hacker News Reader

The Dark Art of Mastering Music

pitchfork.com

133 pointsjonbaer102 comments

This is a long article that starts off stating that mastering is difficult to explain, and then proceeds to not explain mastering.

They're implying that poor mastering was to blame for Metallica's album. Weren't people complaining it was over-compressed? That's mixing not mastering!

Mastering is a matter of making sure that listening to an album from start to finish sounds consistent. That's why songs that go on compilation albums are mastered once on the original album, and again on the compilation - they most certainly aren't mixed again.

Mastering may involve a small amount of compression across all components of the song (although it really shouldn't), but it's mostly controlling loudness, and loudness at different parts of the audio spectrum (more low-end, more high-end). Mastering is effectively something you could do yourself with your stereo's settings (if you were constantly adjusting the settings through-out the song), mixing is fundamentally different and consists of combining "tracks"; vocals, guitar, harmonies, drums, sound effects, adding samples over drums (i.e. mix in a better sounding drum, or hit there-of, over the drumming recorded for the song).

EDIT: For clarity, me stating:

> Mastering may involve a small amount of compression across all components of the song (although it really shouldn't)

is my subjective opinion. Master-bus compression does happen regularly enough during mastering (in addition to during mixing), but as stated, it is typically only a small amount, compared to what a Mixing Engineer can get away with on a per "track" basis.

That's a good explanation of the process and is true in theory/how it should be, but an inexperienced mastering engineer or a clueless band will limit the shit out of a recording. And that could absolutely result in a tinny, over-compressed product. I don't know what happened in Metallica's case, probably failures at different points of the chain.

It's a good point that in general we can't know whether the problem lies (solely) with mastering, because we can't know what state the material was in when it was passed along from the mixing engineer.

Over compression usually happens at the mastering stage, not mixing.

One of the few exceptions is Rush's Vapor Trails, which was recorded so hot there is clipping on some of the instrument tracks.

Clipping is distortion, not compression.

Clipping is both, really. Most distortion is also compression.

And all compression is distortion.

Compression of individual parts happens during mixing, compression of the entire audio track happens during mastering. (Compression of the master bus sometimes also happens during mixing, but mastering engineers prefer not to work with tracks that have already had heavy compression applied to the master bus.)

The kind of brickwall limiting Metallica's album had, where the volume of the track is maxed out at all times, can only come from compressing the master bus, which is properly a part of mastering, not mixing. (See also: the comically overcompressed mix of Raw Power that resulted when Iggy Pop was allowed to remaster the album himself.)

True, that is what I meant by:

> Mastering may involve a small amount of compression across all components of the song (although it really shouldn't)

Admittedly, I've never listened to the Metallica album in question so I don't know what it sounds like. But I thought that in addition to complaints about loudness and lack of variance (compression), there were lots of complaints about distortion (i.e. clipping) etc. Whilst a mixing engineer could theoretically introduce clipping (by pushing the volume beyond its peak) that would be crazy, they typically just do small amounts of master-bus compression to make it sound louder (as you stated).

But yeah, judging by its reputation that album was messed up by several people along the way ;)

That is the opposite of the truth. A small amount of compression across all components of the song is absolutely a legitimate part of mastering; nearly all records have this.

It's excessive compression and limiting that's the problem. Most of the time this only sucks dynamics out of the record, but in extreme cases (like the Metallica record, or the Stooges record) an inept mastering engineer can introducing clipping to the track.

That's subjective. Just because people do something doesn't mean that I can't say they shouldn't do it. I didn't say it never occurs, I specifically said it may occur.

I'm not against master-bus compression, but as it's fundamentally altering the waveform, dynamics of a song and hence artistic expression, then I personally considering it mixing, even if for some reason Mastering Engineers decide to do it.

EDIT: Man, I just realised how stupid that sounds. "master-bus compression" not considered mastering! Okay, perhaps I need to change that perception... ah, nomenclature.

If the mix engineer does it it's mixing.

If the mastering engineer does it it's mastering.

Either can use bus EQ and bus compression.

The mastering engineer usually approaches the job as a final polish of something that's already finished.

The mix engineer can go back and change the mix after adding bus EQ and/or compression. Usually the mix engineer makes [n, which may be a large number depending on time and budget] of mixes, with feedback from the band and producer and sometimes a record company exec.

Then everyone's favourite mix goes off to the mastering engineer for that final polish, assembly into a consistent-sounding album (if it's an album) followed by final format sourcing.

These days artists often do their own mastering. It's not hard to get something better than cheap mastering, but DIY won't be as good the best full-fat pro mastering.

I was always a fan of the sound of the DBX Quantum processor, which is highly rated by professional mastering engineers - not as highly as custom hardware, but a good few steps up from most DIY mastering options.

If you are manually adjusting the volume of the track in response to its level, you are performing compression manually.

The overcompression people complain about in the "loudness war" could have been added at the mixing stage, but is much more likely to have been added at the mastering stage.

> Weren't people complaining it was over-compressed? That's mixing not mastering!

That's not true. Mastering is the process of taking the final output of the master bus and processing it to the final release versions (for the various formats). Depending on where and who does it it can combine many processes: compression, limiting, EQing, stereo widening, bouncing to a tape, etc. It can be a totally functional exercise as well as a creative one.

Something that's over-compressed could have had each instrument on each channel over-compressed (if it was mixed on an SSL for example where there's a compressor per channel), or it could have had far too aggressive master-bus compression. So saying it's definitely mixing not mastering is a stretch.

Legend has it that most of the damage to Death Magnetic was done at the mix stage, not the mastering stage. Master buss compression is absolutely standard practice during mixing. It lends cohesion. Unless you listen to a lot of completely amateur recording, almost all the recorded music you've heard used master buss compression.

On the other hand, it's quite easy for a mix engineer to strap a hard limiter across the buss mix. That's what Metallica did. If the record reaches the mastering engineer with only a couple db of headroom, their hands are tied.

Your post led me down the rabbit hole of over-compression, not for the first time, and I ended up at this video which gave a very succinct answer to my question of: "Okay, I realize that music is getting compressed, but what are we losing?"

https://www.youtube.com/watch?v=3Gmex_4hreQ

Via a nice article with analysis: https://musicmachinery.com/2009/03/23/the-loudness-war/

It's hard getting something that sounds consistently good in all listening environments - I've seen cases where music is offered in digital format either as 'play anywhere' loud .mp3s or 24-bit high dynamic-range .flacs for DJs, but most commercial music is mastered for the poorest environments (in the car, over the radio).

Getting a good master is very important - I'm reminded of a story I heard about an album that was mastered by the artist, and pressed to CD with a high-pitched whine over the top which the artist hadn't been able to hear - like Aphex Twin's Ventolin, but unintentional.

> most commercial music is mastered for the poorest environments (in the car, over the radio).

One thing to consider (at least, that I consider when mastering my own music) is what processing might be applied in the future.

I helped out at the radio station back in college, and you can see the signal chain for audio coming from the studio onto the airwaves. Radio stations often add at least 9 dB of gain via compression, sometimes more. So, if you want a quick and dirty "how does my track sound on the radio?" toss on a master limiter and give yourself 12 dB of gain, then see whether it sounds as you intend (then don't forget to remove the gain after you've made your tweaks).

> a story I heard about an album that was mastered by the artist, and pressed to CD with a high-pitched whine over the top which the artist hadn't been able to hear

Funnily enough, this was common in albums recorded in the 80s and 90s. Reason? CRT monitors were used in the studio, and the high-pitched sound of the screens would get picked up by the microphones. A little discussion here: http://forums.stevehoffman.tv/threads/high-frequency-tones-o...

Certain models of SSL consoles could produce a ~15khz ringing in the master output. I had this exact thing happen to me, and eventhough my hearing good enough to hear it clearly once I was paying attention, in the middle of working I didn't notice it at all, until I got home. I heard that it happened on some Peter Gabriel records (he was famously a fan of SSL consoles). Probably wouldn't be too hard to remove in mastering, incidentally.

Aside from being one of the singers involved, I make archival recordings of my community chorus's performances. In my case, I use a rather simple setup: a pair of small-diaphragm condenser mics (usually but not always set up in the ORTF configuration), a preamp, and a digital recorder. Audacity has all of the capabilities I need for the editing/mastering phase.

One thing I've noticed is that the dynamic range of a live choral performance tends to exceed the useful range of typical consumer audio equipment[1], and it's necessary for me to deal with audience applause (usually by temporary dropping its level), and I also need to carefully apply some limiting on peaks - and sometimes a little bit of compression can work wonders.

If I resort to compression, I typically use parallel ("New York") compression, in which the uncompressed audio gets mixed with the compressed audio. It can be a bit fiddly, since it's necessary to apply the right settings to the compressed track, and to find the right proportion of compressed to uncompressed audio in the final mix.

The finished result tends to sound a bit more "up close" than the raw recording, which can be useful when I'm forced to record from the back of the room.

Even so, I find that I have to turn the volume up quite a bit higher than I would for typical pop music (especially since the Loudness War).

[1]What I mean above is, "being able to listen to the recording at a reasonable level without the peaks distorting or blowing out your speakers."

Actually, the car is one of the best listening environments most people have ready access to - an enclosed space (perhaps dolby 5.1) with sufficient bass, with all the channels pointed more or less equally at your head. Many audio engineers use a car as a reference point.

I dunno. Between road noise and my crappy speakers, GP made perfect sense to me. :)

As a bedroom producer, I have to add that good mixing (carefully tuning each audio channel to remove frequencies that interfere with another channel) is more important than mastering. Mastering adds the finishing touches, but can't really fix mixing errors.

Also don't get fooled by opinions like this:

“When you only use computer stuff, everything starts smearing together, and a whole other mechanical aspect that isn’t musical creeps in. But the better the quality of your gear, the more the thing will sound real.”

Obviously high end gear sounds great, but that doesn't mean that digital plugins and emulations can't sound equally good.

The only part where I see an advantage on the whole audio - digital debate is modulation, but that's a different stage of making music.

bedroom producer here too ! the greatest thing about making music all in the box (or majority of it) is being able to mix on the go.

majority of my tracks sound great (imo) before I even touch a limiter/mastering chain.

lots of advice online says 'make the track and mix after' which is good if you are in the flow and can bust out a track in a day.. but if I make a synth sound I immediately mix it into the song and then start playing with chords

I also blended it into the workflow. It's really fast to add a high pass filter, set the gain and remove artifacts if needed. Part of sound design is mixing also for me.

Before exporting I revisit all channels, but usually not much work left. It probably helps a lot that I do electronic music mostly.

Everyone works differently. My problem is that if I start mixing as soon as I introduce a new synth, I get really obsessive and perfectionist over it and lose the big picture.

At the moment, I try to just do some light sound design and mixing while building the track, then do the full mixing at the end.

> the greatest thing about making music all in the box (or majority of it) is being able to mix on the go.

That's no different to working out-of-the box. I have a 32 channel analogue mixing desk, many analogue synths and drum machines, outboard effects units (Lexicon reverb, 3x Eventides), outboard compressors (2x 1176s, SSL G buss, 8 x SSL 9K channel comps, DBX 1066), etc.

The experience is much more like performing the studio, rather than tweaking something with a mouse one knob at a time. I find sounds, play something live with the track, get the levels right on the desk, EQ the channel on the desk, patch in compressors to shape the sound, and have 6 auxes always setup for reverbs, delays, etc. I can get to the sound I want much quicker than I ever could before in-the-box.

Obviously it's horses for courses. Whatever works for you, it's the end result that counts. But being able to mix whilst making the music isn't exclusive to in-the-box.

> Obviously high end gear sounds great, but that doesn't mean that digital plugins and emulations can't sound equally good.

As a producer that started in-the-box and built out into pretty much 100% outboard gear (mostly analogue), I agree with you in principle. But to get something that sounds 'equally as good' definitely takes a lot more effort with plugins. For me personally this is a big deal because I thrive on being able to realise my ideas quickly - both in musical terms and sound design terms.

Relevant: the Dynamic Range database

http://dr.loudness-war.info/

This is great, thanks.

Handy write-up on what "dynamic range" means in practice here: https://www.soundonsound.com/sos/sep11/articles/loudness.htm

This is just a one-time reminder that this whole magilla is about one kind of music: the recorded, mass-produced consumer object.

Although it may seem large and important due to its ubiquity and market value, there's really much more to music than this. Music can sound good and be valuable 100% live with no mics, outdoors with no special architecture. The mastering engineering in the article espouses the virtue of unique equipment, but in a live band, nearly every instrument and every performance quickly becomes a unique one.

Ask yourself, who does it really serve, to experience music as a lithographic impression rather than a unique experience? I'm not even arguing about musician's pay or anything like that. I'm talking about qualia of richness: diversity, local character, dynamic range, interplay with your mood, involvement, movement.

My comment here is really to just break up the spontaneous hierarchy of dollars + equipment + mystery -> ideal studio product. Now that technology has empowered us all to make more music faster on our own I feel it's especially important to ditch the last century's mindsets of scarcity and monoculture with respect to music.

"Ask yourself, who does it really serve, to experience music as a lithographic impression rather than a unique experience?"

A recording allows me to share a specific performance with someone else. It's like a painting, but in audio. I can tell someone how much I like the song Heart Shaped Box on Nirvana's In Utero album, and introduce them to that particular version I like. If my only option is to introduce them to the song at a concert, then they might end up with something like the version on Muddy Banks Of The Wishkah, which is a different experience. (In my opinion, an inferior one.)

Live & unique performance of music has value too, and it's nice to have those fleeting moments that are only shared with the people who were with you at the time. But recording can be a great way to appreciate and enjoy the art itself.

I seldom see such hippie bullshit espoused on HN outside of Bitcoin/Ethereum threads.

Music can sound good and be valuable in any setting: it can sound better with good audio engineering.

It serves everyone to have a high quality artifact.

> technology has empowered us all to make more music faster on our own

That's not what I want; that's not what any of us wants.

I would bet a few shekels that the parent just finished reading the recent "Records Ruin The Landscape" by David Grubbs and is making it count on the internet.

I hadn't heard of that until you mentioned it. I'm familiar with many of the works listed at [1], but considering the reported theses for a moment "new genres in experimental and avant-garde music in the 1960s actively thwarted the form of the LP" and

"listeners coming to know a period through the recorded artifacts of composers and musicians that largely disavowed recordings." [1] I think is interesting and ironic. But, my argument isn't about the mismatch of recording technology and artistic activity, but of mass reproduction and especially the optimization of recordings for profit.

Even without creative performances such as the Fluxus movements and sound artists of that era, pop music can flourish without massive lithography. The beautiful thing is it perpetuates itself through continuous reinterpretation by musicians and listeners, resulting in an accelerated, rich evolution. That's how we got dance steps like the Tango.

[1] https://dukeupress.wordpress.com/2014/03/21/records-ruin-the...

> such hippie bullshit

This breaks the HN guidelines, which ask you not to call names in arguments:

https://news.ycombinator.com/newsguidelines.html

The comment would be much more substantive without the first sentence. Please edit such stuff out of your posts here.

There are lots of musicians for whom the studio is their instrument. Arguably, everything the Beatles did post 'Revolver' wasn't even possible in a live setting. There's certainly something to be said for the organic interplay of live musicians, but making consumer artifacts isn't necessarily 'bad' – Warhol built his entire career on making art out of them.

There are online mastering services and the results are tragic. Take a couple devs who are into music and maybe have some math chops but no actual mixing/mastering experience and no access to real equipment either. A little open source software and some headphones will do. Now throw in a music industry "endorser" (these guys are always investing in the wrong companies!) and spew out some automated schlock.

http://promasterhd.com/ http://landr.com

"Pro" this and "industry" that. The desperation is palpable. And often plainly visible in their faces:

http://blog.masterlizer.com/index.php/2015/12/14/re-imagine-...

Regardless people always prefer "the louder one" so you can immediately see where this is headed. Generate a "mastered" .wav file using open source tools and a couple Python rules, convert it to MP3 without paying the license fee, then charge people $9 if they want the "higher quality" .wav. $14 if they want all 24 bits instead of you running it through ffmpeg a second time.

Meanwhile SoundCloud streams 128k MP3s like we're stuck in 1992. Imagine if Vimeo or Flickr limited stuff to an equivalent lossy compression quality. People would not stand for it. But with audio it's somehow okay.

Now all we have to do is combine the two. Say... robo-mastered junk being served up as a SoundCloud MP3. Coming soon!

>Meanwhile SoundCloud streams 128k MP3s like we're stuck in 1992. Imagine if Vimeo or Flickr limited stuff to an equivalent lossy compression quality. People would not stand for it. But with audio it's somehow okay

HD screens are pretty much ubiquitous. Headphones (or speakers) that aren't trash I would guess less so.

Also, humans prioritize and are sensitive to visual input more than sound to a large degree.

I'm sure dogmen would have excellent audio systems and have perfected HD smell-o-vision alongside black and white VGA monitors.

> Also, humans prioritize and are sensitive to visual input more than sound to a large degree.

Our evolved ability to detect and localize sound in the dark at night is why we're here typing. Ironically codecs tend to see phase information and say "hey, we can ditch most of that and save a few bits!"

Any filmmaker will tell you that audio can make or break a picture. But for some reason audio always gets screwed commercially -- especially among technologists. The early iPod headphones and even modern MacBook/iPad audio is awful. The iPhone still only has a mono microphone. Beats headphones are shameful. Satellite radio is godawful. Anyone here want to sing the praises of "RealAudio"? Yet people tolerate it because there's no industry leader standing up for quality. And even when someone does, they're usually off in the weeds themselves (192k audio fad).

SoundCloud is a hopeless business, bleeding cash. Fred Wilson can claim he loves music but apparently he's not willing to say "Hey, maybe y'all should serve something that sounds at least as good as iTunes did in 2001." Like no other artistic medium I'm aware of, music appears to be worth significantly less than the bandwidth required to stream it properly.

The point of services like LANDR is that good mastering requires you either become very skilled and experienced, or spend a lot of money. Their output isn't great, but it many cases it's better than no mastering at all. Also, I imagine the technology will eventually improve.

I agree Soundcloud's shitty low bit rate MP3s are totally unacceptable, though. I never use Soundcloud for actual music listening.

I suppose. But pushing out a low-quality version of a previously high-quality service isn't disruption; it's just peeing in the pool. LANDR is now on version three of their algorithm and it still sucks. The goal here is to move people emotionally and maybe re-light the spark of a dying industry. This is the wrong direction:

"'There are good and bad things about it,' mastering professional Dave Gardner (Louis C.K., the Hold Steady, Black Lips) told Billboard in January. 'If someone can spend $10 a month and get something better than what they have but not very good, I don't know if that's going to make people appreciate mastering even more, or if it's the nail in the coffin.'"

Still, I don't think LANDR is intended for people who can regularly afford professional mastering. It's like saying netbooks are going to devalue the Macbook.

>LANDR is now on version three of their algorithm and it still sucks.

Sure, but can you imagine how difficult it must be to implement good automatic mastering? It may be a decade or so until it comes close, but eventually, I bet it'll become a viable option for many bedroom producers.

A neat read I think, even touches on LANDR, a new-ish and pretty compelling service. I've only tried it once but it worked as intended - tweaked a file, made a difference. Improved? Too subjective, but a tentative yes.

I do take such an article with a grain of salt though, because a lot of the "tightest" sounding music since 2010 - in my PERSONAL opinion - is in drum & bass and, more than likely, not run through too much outboard gear like is romanticized here. Smart routing in Ableton and into iZotope isn't outside the reach of intelligent, determined artists and producers of many stripes.

I just tend to think "Oh, yeah Grimes sounds pretty good, so does Radiohead, but compared to what's happening in a Netsky or Camo & Krooked track, just different ballparks altogether." Like I said, that's me and my ears. Having great outboard gear can truly make a difference if the source material is great.

> Traditionally, the “marketplace” has been radio, where a well-mastered song hits that sweet spot where you feel immersed in the music but not battered by it. If your song is poorly mastered, the logic goes, people won’t want to buy your album.

Part of the problem here is that many radio stations will run incredibly aggressive limiters/maximisers across their output bus anyway, for the exact same reasons given in the following paragraph. Your average pop track is being compressed to hell and back at three stages: mixing (gotta make that synth subgroup 'pop'), mastering (gotta be louder than my rival pop-group) and transmission (gotta be louder than my rival station).

To be fair, almost all cases where you're listening to FM radio, you're listening on a crappy device.

I remember having Sirius in two of my previous vehicles and not noticing it being that poor of sound quality from low digital compression rates, compared to CD's or what I'd get from my iTunes on my phone. However, in my current vehicle which has a pretty solid factory system, the digital artifacts are exceptionally glaring.

When you're on a bad system, some of the compression or limiting happening to the track is either a positive or not noticeable.

Have heard some arguments that the UK's (world's?) digital protocol, DAB is actually a worse experience for listeners than a decent FM station, but I'm not all that well informed on that.

Quick clarification on digital versus the kind of compression discussed in the article: with the former you're making the file size smaller and losing a little bit of quality in the process (sometimes leaving artifacts), whereas with the latter you are seeking to make the loud bits quieter, meaning you can make the whole lot louder without distortion.

Another thing worth noting is that for the vast majority of users, the car will be the best listening environment they are exposed to. When I had a very minor foray into the recording world, I remember the mix engineers trying out bouncedowns around the car park.

In my opinion, with Sirius when the digital compression is noticeable, it's a worse listening experience than normal FM analog compression and fidelity, so I'd imagine that's true for other over-the-air digital compression.

I was surprised to find that out myself. I bought a used car a year ago with satellite radio hardware, but it didn't come with a subscription. Shortly after I got it, Sirius did a promotion where it was free for anyone with a receiver for a few days. I couldn't believe how bad it sounded: substantially worse than FM, and actively fatiguing and unpleasant to listen to. Despite the range of decent stations offered (and my hatred of commercials) it was bad enough to put me off signing up.

Interestingly, I thought the always-free "Preview" station was of noticeably higher quality.

That's mostly wrong. DAB uses 128kbps or 192kbps MPEG2 audio, which is pretty similar to MP3 at those bitrates; either is way better than FM radio, except for material that breaks the algorithm (noise, applause of some sorts).

...and cymbals? That's always my go-to tell.

Cymbals also tend to suffer very quickly with strong dynamics compression. The most recent Prodigy album The Day is my Enemy has a lot of that, unfortunately. I much prefer their 2nd, Music for the Jilted Generation :)

I'm not sure which of the parent posters started but it seems like we're conflating two rather unrelated meanings of the word "compression" here: compression in the sense of data-compression (which is often lossy for many audio formats like mp3) and compression in the sense of dynamics-compression, which is what is meant by "compression" in the context of mastering.

Most of the music channels at Sirius (when i was in a position to have knowledge) were being broadcast at 32 kb/s. I wouldn't call myself a hard core audiophile, but that quality is below the threshold of what I enjoy listening to.

Good headphones are essential. I've had a number of tracks/albums where I've listened to them on my computer with nice headphones, then gone to the gym with my portable headphones or driven home listening over the speakers and the songs sound terrible.

I've often wondered if the rise of some genres has been due to shitty mp3 quality and even shittier earbuds.

Very true. I've also had the opposite experience. Bad speakers/headphones can hide flaws. On good headphones, sibilant frequencies, compression artifacts, and muddy bass can all become really apparent and annoying.

Mastering is like buying good wine. There's a lot return for your buck at the low and medium end. The returns begin diminishing steeply in the high end. Above some point, it becomes purely a game of audio engineers impressing other audio engineers.

Like pretty much most things.

Haha yep! At least in subjective realms; maybe slightly less so for practical realms.

I see the "loudness wars" are still raging. Way back when I was a recording engineer (early 2000's) I remember reading about it for the first time. I had this "a ha" movement. Until then, I couldn't put my finger on why I got fatigued listening to certain albums: even one's where I liked the content of the songs.

Now if a record is over-compressed I don't even bother(sorry Foo Fighters, get your shit together). Also, stay away from "remastered" version of old albums. That really means they've over compressed it and killed all of the dynamic range of an album you previously loved.

Mastering is a subject about which a great deal of mumbo jumbo is talked, but a very important point that this article does put across well is the necessity of making sure the music sounds good across a wide variety of playback environments. Unfortunately, this kind of "optimizing for all cases" comes at the expense of how good the music can sound in any one environment.

(For example, a lot of people will have heard recordings that produce incredibly detailed spacial perception, but only when listened to on headphones - and which sound pretty poor on loudspeakers.)

We got into this situation because of the economies of mass produced physical media, but we largely don't have those constraints any more. Since the cost of digital distribution is basically a rounding error, we could be producing different releases optimized for different settings and letting the listener choose which is the most appropriate for them.

> Unfortunately, this kind of "optimizing for all cases" comes at the expense of how good the music can sound in any one environment.

Sorry, but that's not the case. The reason music gets mastered poorly today is because of poor mastering engineers. I've had this conversation numerous times with an engineer that has designed some of the most well-received studio equipment (including analog compressors) ever made.

Listen to The Dark Side of the Moon. Regardless of your like or dislike of Pink Floyd that album sounds good regardless of how you're playing it (car stereo, single speaker boom box, headphones).

Now why there aren't many good mastering engineers left (or properly equipped studios) is a separate, lengthy conversation.

I'm certainly not claiming it's the reason why music gets mastered poorly today, only that things like representation of space, dynamic range, audibility of fine detail vs. making the music sound too cluttered, are and always have been trade-offs.

As far as why things are worse lately, mastering engineers are working for hire. If people think they're producing bad work, it's probably at their client's behest.

Sounding good on many systems doesn't mean it couldn't have sounded better on any particular system had it been mastered with that particular setup in mind.

The idea that there is some magic mastering ability that can overcome the pitfalls of the vast array of listening environments and equipment is exactly the type mumbojumbo the OP mentions. The audio world depends on it.

I don't think today's professional mastering engineers are "poor" necessarily; some of this I think is marketing pressure. The Loudness War (https://en.wikipedia.org/wiki/Loudness_war) is a real thing and a large part of the decrease in modern mastering quality, in my opinion.

The thing is, this mastering technique has some upsides for both casual listening (it will stand out more from the rest of the pack) and also will sound "louder" on poor equipment without necessarily exceeding the equipment's capabilities. The significant downside is that a lot of the detail is lost at best, and at worst you get very audible clipping / distortion or unnatural "pumping" effects. So not good at all for those that go deeper in their music.

"Optimizing for all cases" might help end this loudness war though. Many of the online streaming services (Youtube, Spotify, iTunes, etc.) have some optimizing routines that aim for consistency in volume level. The net result is that extremely over-compressed music sound dull and flaccid.

Many articles recommend more sensible overall loudness levels now... although I haven't seen a specific LUFS number to hit, like there is for European television (EBU R128), aiming for something like -16LUFS as this article mentions is a much better situation than before.

See: https://www.soundonsound.com/sos/feb14/articles/loudness-war...

I think you are wrong. Due to the rise of the bedroom producer and available software we've never before had so many talented sound engineers. The production value on electronic music today is through the roof.

Also, due to the rise of the bedroom producer and available software we've never before had so many UNtalented sound engineers.

Of course, but the competition is fierce.

I always tell my clients that mastering is the final step in mixing: it's the part where the mix engineer gives the mix to a different engineer in a different studio with a different system and different ears and different gear and different opinions to put the final touches on the 2-track mix. The tools are almost always various flavors of EQ and compression, but the most important part is all the "different" stuff.

The loudness wars of the late 1990s and continuing somewhat today I believe are a relic of CD (and FM) audio and will pass now that we have ubiquitous volume leveling playback devices.

It's well understood in audio (and demonstrable in ABX testing) that "louder sounds better." If your CD is mastered a little louder than the other guys, then when you change it (or when your CD jukebox changes it) then your CD will "pop out" more, be heard better, and sound better.

Enter volume leveling.

When the playback device matches the playback volume of two songs, the one which is overcompressed will sound quieter and yet be more fatiguing, while the one which has better dynamics will actually feel louder and punchier yet be less fatiguing. Volume leveling actually stands "louder sounds better" on its head: louder will not sound better. In a strange way, iTunes / Replay Gain is helping to bring back audiophile mastering into vogue. And you can tell - mastering has definitely changed for the better in the last five years. There's a lot more attention to dynamics, especially with artists learning about having to master their digital audio for vinyl, and how that changes things.

I wrote about the loudness war in 2002 in "Over the Limit"[0] and about the end of the loudness war in 2009 in "Over the Limit 2"[1]. In OtL2 I predicted it would be another decade - until 2019 - before the loudness war finally could be declared over.

From what I've seen I think I'm on track to hit my prediction. I'm glad.

[0] http://riprowan.com/over-the-limit/

[1] http://riprowan.com/over-the-limit-2/

For a bedroom producer/DJ, LANDR is a godsend. When you have 10 2-minute drafts, each of which was complete in an hour or two, and want to take them for a test-drive in your next gig, before LANDR you had to spend an additional 20-30 minutes per each of them, to get them into a reasonably compressed and loud shape, so you won't have to desperately ride levels and eq from track to track in front of the crowd. And $15 per month for unlimited demos and drafts is a very sweet price point compared to even the cheapest mastering engineers.

Sure, but you get what you pay for. In both cases.

Couldn't you do much of that just by using presets in mastering software like Izotope Ozone 7? I don't really understand the benefit LANDR offers, unless there really is some secret sauce in the settings LANDR chooses automatically. But if that's the case, I imagine Ozone will features similar functionality soon.

First of all, Ozone costs a lot more. Second, every time you're using a tool with a lot of settings you find yourself reaching to tweak them — which is one of the reasons why in making music it's often better to use minimalistic tools that don't offer too many options. And finally, judging by the results I'm getting, LANDR is more "intellectual" that Ozone and tweaks it's settings depending on many properties of the source track.

> Professional audio engineers operate out of carefully calibrated rooms designed to eradicate odd resonances and echoes, and their speakers cost more than your car.

That is not exactly true. I've seen many pictures from studios and most engineers use professional speakers (called studio monitors) in the $5K range.

However you can spend a lot more on acoustic treatments, they make a massive difference.

I like to think that color-grading, another mysterious dark art (to me at least), is the visual equivalent of mastering.

Pitchfork is not a publication that is targeted at software engineers.

Decent nearfields cost $4-5k a pair. A proper pair of main monitors from the likes of PMC or Quested cost $50k to $150k.

It's easy to overlook the main monitors, because they tend to be recessed in the walls.

http://vintageking.com/pmc-loudspeakers-qb1-pair http://vintageking.com/quested-hm415-active

Define decent. You can buy a pair of Mackie from the MR series from about $500. That good enough for a long period till you learn what you are doing.

Audio gear has the tendency that the value you get with more bucks decreases exponentially.

If you haven't learned what you're doing, you're not the professional audio engineer referred to by TFA :)

That's true, all I'm saying is that you don't have to pop $4k on a monitor speaker, you can buy decent ones for 1/10th of the price.

I personally think we should be able to buy music in its pre-mastered form. Yes, that will be a lot of data, but nowadays that is not really an issue anymore. The experience will be so much better than our current option, which is basically only the EQ settings on our amps.

It would be nice for me to get individual master tracks for sampling purposes. I would pay for access too, but I've never even seen that as an option.

This is exactly what the Stems music format is all about: http://www.stems-music.com

Stems is only four tracks. That certainly works well for "standardized contemporary music for DJ" but it doesn't sound it is a universal format for all kind of music.

Might just be a terminology thing, but I don't think mastering engineers work off of tracks. I think they're given mixdowns and adjust from there.

There are some fascinating isolated tracks out there, but I wonder if knowing that stems will get released will push things in the studio even deeper into the territory that every recorded note must be perfect. Making a record already produces a lot of self conscious anxiety that affects your decision making, so knowing that people will also be listening to and analyzing your individual tracks is even worse!

That's something that indie bands could consider, and would be an easy addition to a Super Deluxe fan edition of a release.

I also wish more bands would release instrumental mixes of their albums, like Imogen Heap, Celldweller & Blue Stahli have done. Sometimes I want to listen to an album while I work but the vocals are distracting. I've been happy to pay a few dollars extra to get the instrumental versions when available, it's an easy way for indie bands to make some extra money from fans.

You'd prefer no mastering over well-done mastering?

So, from the comments here, it seems that people are under the impression that mastering engineers work off of the individual instrument tracks (the stems). This is not true. The article itself explains that a mixing engineer takes the individual tracks and mixes them. The final mixdown is then sent to the mastering engineer. It's called "mastering" because this stage of processing happens on a single track - the master track. Buying music in a pre-mastered form would still be a single lossless file, which is no different from Bandcamp allowing you to download the final lossless mastered version. The pre-mastered tracks will just sound worse and have wildly different volume levels depending on how they're normalized.

If you want the actual stems of the song, that would actually be the music in it's pre-mixed form. It wouldn't even be mixed at that point, let alone mastered.

Yeah, there seem to be very definitive statements in these comments without a basic understanding of the recording process. I wonder if these are the same people who swear they can tell the difference between various playback formats or when they use gold XLR cables.

Apart from the misunderstanding, I think it would be actually interesting if you could buy the un-mastered version, plus "mastering" files to use depending on your setup (e.g.: crappy headphones, car stereo, home stereo), but that would require multiple mastering sessions and wouldn't probably be very cost effective...

Or you could crowd-source them but then it might go against the creators intentions...

In a different field (digital cinema) the above is being done for color, lookup the ACES colorspace and workflows...

(Very) Basically the camera's recording goes through a process to make it into a standardized ACES colorspace. It will then being processed by editing, VFX, grading, etc... and then there's output transforms from ACES for each output device, like a cinema projector, a home TV or a mobile phone.

And yes, AFAIR, that does need grading for each targeted output device, because of how we see and perceive color depending on the environment we're in...

It would be awesome (EDIT: to have something like that) for music as well

I don't understand how unmastered tracks would be a lot of data. Can you explain? Also, I cannot imagine wanting to mess around with EQs for each track on an entire unmastered album. That sounds like a huge burden.

Maybe he meant pre-mixing instead of pre-mastering, ie - having all the individual tracks. But that would require special playback software, more akin to a DAW.

Even given that misconception, I don't understand how that would equate to a better listening experience.

There are some songs where I hate a particular instrument, or I would like another one to be louder. So I can see why this would appeal to some (limited) group.

What you want is a kind of multi-track audio format.

Some exist: http://news.traxsource.com/articles/1754/stems-announced-at-... only four tracks dedicated to designated instruments. I've also heard of iklax http://www.iklax.com/

It would be fun for other recording professionals and hobbyists, to deconstruct a record, and maybe some super fans, but overall it seems like something people wouldn't care that much about.

But how granular would that get? Would I be able to listen to one of 50+ layered distorted guitar tracks on "Siamese Dream"? Would anyone want to do that?

A few years ago Radiohead released the high-level stems for Reckoner (man, I guess that is going on 9 or so years now!). They let people download them and create their own remixes of the song. It was a lot of fun, and I remember listening to some pretty cool remixes.

There's a good article from 2001 which explains this in depth: http://www.cdmasteringservices.com/dynamicrange.htm

Edit: Whoops wrong thread. :)

I think you might've posted in the wrong thread, and actually meant to post here instead: https://news.ycombinator.com/item?id=11754414

Thank you for catching that.

Reply on news.ycombinator.com