Why Mastering Matters More Than Format
I want to make an argument that should be obvious to anyone who has spent time around recording studios but somehow never gets made clearly a lot of the time in audiophile circles: the mastering of a recording matters more than the format it’s delivered in. A well-mastered CD will beat a poorly-mastered 24-bit/96kHz download every time. A 1987 pressing of an album can sound better than the 2024 hi-res remaster of the same album. The format is the container. What’s in the container is what you’re actually listening to.
This isn’t a subtle point. But it gets lost in a hobby that spends enormous energy arguing about speakers, amps, sample rates, bit depths, and codec comparisons while ignoring the thing that actually determines how music sounds: the decisions made by the mastering engineer before the file ever reached your hard drive.
I was a mixing engineer. I’ve spent years in rooms where these decisions get made and I know what they sound like when they go wrong. This is the post I’ve been putting off writing because it steps on some comfortable assumptions about what hi-res audio means.
What Mastering Actually Does
Mastering is the final stage of audio production before a recording is released. The mastering engineer takes the completed mix and prepares it for distribution adjusting the overall tonal balance, setting the final loudness level, ensuring the stereo image is correct, and sequencing the album if there are multiple tracks. On a good day, mastering is largely invisible: it makes the mix sound like the best version of itself without calling attention to anything.
The main tool in a mastering engineer’s kit for controlling loudness is a limiter. A limiter is a type of compressor that sets a ceiling. Audio above that ceiling gets attenuated, pushed back down. Used with restraint, limiting is how you ensure a recording doesn’t clip (distort) during playback and how you give it the right amount of overall energy. Used aggressively, it’s how you destroy a record.
When you apply heavy limiting to a mix, a few things happen. The loudest peaks like the attack of a snare drum, the pick of a guitar string, the initial consonant of a vocal get pushed down relative to everything else. The music gets louder overall but it gets flatter. The dynamic difference between the loud parts and the quiet parts shrinks. The music stops breathing. This is what people mean when they say a recording sounds “squashed” or “fatigued” the term is correct, it’s what heavy limiting does to a waveform.
If you push limiting far enough, you get clipping. Clipping is when the audio signal exceeds the maximum level the system can represent and the top of the waveform gets cut off and literally clipped. Digital clipping sounds like distortion. It’s audible as a harshness in the high frequencies and a blurring of detail that’s not always immediately obvious but becomes fatiguing over the length of an album. Many commercially released recordings from the 2000s have clipped audio. You own some of them.
The Loudness War and What It Did
From roughly the mid-1990s through the early 2010s, commercial recordings got progressively louder. Not because anyone decided louder was better, but because of competitive pressure: radio programmers preferred tracks that sounded louder in the context of a playlist, so labels pushed mastering engineers to make masters louder, so the next release had to be louder to compete, so the one after that had to be louder still. The result was an arms race measured in dynamic range, the difference between the average level of a recording and its peak level that ended up compressing that range toward zero on a lot of commercially successful records.
The casualties are specific and well-documented. Metallica’s Death Magnetic (2008) had such aggressively limited masters that the Guitar Hero version of the same album derived from the pre-mastering stems sounded audibly better. The band acknowledged this was a problem. The 2023 remaster corrected it: more dynamic range, no clipping, the same performances but finally with room to breathe. If you have the original CD pressing of Death Magnetic and you haven’t heard the remaster, they are different listening experiences despite being the same format.
Red Hot Chili Peppers’ Californication (1999) is another well-known case. The album contains confirmed digital clipping across multiple tracks audio that exceeded 0dBFS and got cut off at the top. This is not a subtle audiophile concern. It’s measurable distortion on a commercially released major label record that sold fifteen million copies. The vinyl pressing sounds different because the lathe cutting process naturally limits how loud vinyl can be cut, so the vinyl master had less limiting applied than the CD.
Streaming normalization has helped more recent releases. Spotify, Apple Music, and Tidal all normalize playback loudness, which removes the competitive advantage of a loud master. If your record gets turned down to match everything else on the platform, there’s no point in crushing the dynamics to make it louder. The loudness war has receded on streaming platforms as a result. But the catalog, everything released during the peak loudness war years is still out there in its original state, and a lot of remasters of that material have continued the same habits.
Why Some 1980s CDs Sound Better Than Modern Hi-Res Remasters
This is the uncomfortable part of the argument and the part I want to be specific about, because it sounds like audiophile mysticism and it isn’t.
Early CD pressings particularly Japanese pressings from the 1980s were mastered at a time when engineers were still figuring out what CD could do. The default approach was conservative: don’t push the levels too hard, preserve the dynamic range of the original tape, don’t do anything you wouldn’t do on vinyl. The result was CDs that often sound remarkably dynamic and detailed, not because of any magic in the format but because the mastering decisions were more restrained.
Then remasters happened. A remaster is supposed to be a transfer of the original recording to a new medium, ideally with improved quality from better equipment and better understanding of the source material. Sometimes that’s what you get. More often you get a mastering engineer who has been asked by a label to make the remaster sound “competitive” with current releases, which means louder, which means more limiting.
The result is remasters that are technically hi-res 24-bit, 96kHz, higher resolution than the original CD, but that have less dynamic range, more limiting artifacts, and in some cases clipping that wasn’t in the original pressing. You’re getting more bits per sample describing a worse-sounding signal.
Specific examples that come up repeatedly in mastering discussions: the 2011 remaster of the Dark Side of the Moon stereo version is brighter and more limited than the original 1983 CD. The 1980s pressings of many Led Zeppelin albums have measurably more dynamic range than some of their subsequent remasters. The original Japanese pressings of certain 1970s rock albums trade for significant money on the used market specifically because the mastering is better, not because there’s anything magical about Japan.
I want to be clear: not all remasters are worse. The 2023 Dark Side of the Moon 50th anniversary Atmos Blu-ray is genuinely better than the original CD in every way that matters. It’s a new mix, more dynamic range, no limiting artifacts, with Atmos spatial dimension on top of it. The Beatles remasters from 2009 are careful, well-executed transfers. The problem isn’t remastering. It’s remastering done for competitive loudness rather than fidelity.
How to Tell if a Recording Has Been Damaged
There are tools that make this measurable rather than subjective.
Dynamic Range Database (dr.loudness-war.info). The DR Database is a community-compiled database of dynamic range measurements for commercially released albums across multiple pressings. It uses the DR measurement standard a simplified but useful metric that measures the difference between average and peak levels. A DR of 14 or above generally indicates a recording with healthy dynamic range. DR of 8 or below is where you start to hear the effects of heavy limiting. DR of 5 or below is aggressively limited. Californication measures DR5 on CD. Original pressings of many 1970s and 1980s albums measure DR12-16. The difference is audible.
The database lets you compare multiple pressings of the same album. If the 1987 Japanese CD of an album measures DR13 and the 2009 remaster measures DR8, the original pressing has more dynamic range by a measurable amount. Whether that matters to you depends on the music a classical recording at DR8 sounds worse than the same recording at DR13 more noticeably than a punk record does, because classical music relies on dynamic contrast in a way that punk doesn’t.
Spectrogram analysis. A spectrogram shows the frequency content of audio over time. Clipped audio has a characteristic look. The waveform hits the ceiling and the top gets cut off, visible as a flat line at the maximum level. Software like Audacity (free) or Adobe Audition can show you this. If you suspect a recording has clipping, look at the waveform in an audio editor. Flat tops are not normal. Flat tops are distortion.
Your ears. Heavy limiting has a characteristic sound that becomes easier to identify once you know what you’re listening for. The attack of percussive sounds like drums, piano, acoustic guitar sound softened or blunted. Sustained notes feel compressed. The music doesn’t have peaks and valleys; it’s a consistent wall of sound. After a while it’s tiring to listen to, in the same way a conversation where someone speaks at the same volume and pace for an hour is tiring. Music is supposed to have dynamics. When it doesn’t, something’s wrong.
What This Means for Buying Hi-Res
Hi-res audio is not automatically better audio. A 24-bit/192kHz file of a badly mastered recording is a high-resolution capture of a badly mastered recording. The extra bits and the higher sample rate are describing something you wouldn’t want to hear in any resolution.
Before buying a hi-res download or a remastered disc, it’s worth spending five minutes checking the DR Database for that specific release. If the remaster measures significantly lower than the original pressing, the original pressing is usually the better buy even if it’s technically lower resolution. Sound quality isn’t determined by the number after the slash in “24/96.” It’s determined by the decisions made by the people who made the recording and the people who prepared it for release.
The reason this blog exists is because lossless extraction from disc gives you the best possible version of what’s on that disc. But “best possible version of what’s on the disc” is only meaningful if what’s on the disc is worth preserving. The format question and the mastering question are separate questions. Both matter. The mastering question matters more.
The streaming vs disc mastering post covers a related angle the difference between the master used for streaming and the master on the disc, which is a separate problem from the loudness war but often compounds it. The Dark Side of the Moon formats post has a worked example of what the mastering differences between versions of a specific album actually sound like in practice.