I’m intrigued by this week’s Neil Young comment that “Steve Jobs preferred vinyl” when he listened to music, and we need a new higher quality digital music format.
I was just talking about this stuff with an audio engineer a couple weeks ago. We had a great conversation about the Rick Rubin “louder is better” mentality, the way a lot of music is mastered today.
There are also the realities of our digital age: part of this discussion is file size, not just as related to storage on the device, but also in the context of the time it takes to download a song and the server and network power required to store and manage that download or stream.
Is Neil Young just way too late here? Will the average listener care about quality, or does the use case for compression win out?
Vinyl is analog. Audio is presented in its full quality, and people spent lives trying to create, record, and master the richness of sound and get it on a record. Complex sounds sometimes presented themselves with minor inexactness and even outright flaws, but that gave the sound a real, human quality.
Digital technology began to evolve in the 70s, and it began encroaching on music. If you grab a Queen album, you’ll see on the liner notes the cry “No Synthesizers!”, which is a direct dig at one of their contemporaries – ELO. But the move was underway.
CD technology debuted on its first major album, Billy Joel’s “52nd Street” rereleased from its 1979 original vinyl version, in 1982 along with Sony’s first full production CD player. The reduction in quality had begun, but the use case for digital began to take hold. CDs use Reed-Solomon encoding, which I’ll always have fond memories of because I was lucky enough to sit in a class taught by Irving Reed for a semester. The plus side is that CIRC avoids complete failure – the dread skip – resulting from a good size surface scratch. Along with the sheer size and portability difference, this was a much more important feature to users than any perceived difference in sound quality. The debate over CD versus vinyl quality is very technical, as long as one stays out of “clipping” the dynamic range. More shortly.
Then the digital recording age came, with Yes “Talk” being one of the first major albums run straight to 10 GB of hard disk (on four Macs) in 1994. This fully captured one major advantage digital technology had: sound could be recreated, exactly as mastered, indefinitely. The same benefit that applied to CDs surviving scratches from handling now applied to studio production – masters didn’t degrade over time and use. Free of the restrictions of width and depth of physical grooves and crystal needles, loudness suddenly isn’t a problem either.
Coincidentally, MP3 was released as a standard in 1993, and the compression era started. The folks at Fraunhofer applied a lot of psychoacoustic trickery, and a bit of Suzanne Vega, to come up with a format that sounded like the real thing in 1/11 of the space. It’s an inherently lossy algorithm by design, but by comparison to the loss in dynamic range over a format like FM radio, it was acceptable to the vast majority of ears.
That launched the wave of digital music download and mobile players, which created not only behavior change in users, but in producers, labels, and artists. Since I can’t play you the songs (*smirks*), we’ll try some visual aids to explain what happened in the evolution of one band in the middle of all this.
From 1986, Metallica’s “Welcome Home (Sanitarium)” visualized from its MP3 file in Audacity. It’s one of the better sounding songs from their earlier somewhat self-produced days. Like a lot of good Metallica compositions, the first minute is rather quiet and all instrumental. Notice how much dynamic range it uses – you can actually see the sound build into where things get loud at 1:50. There’s very little clipping, where you see the sound maxed out. Also, the difference between the left and right channels are noticeable even on this scale.
Bob Rock joins as producer in 1991, and tells the band they can “sound better” recorded. The Black Album gives us “The Unforgiven”, with a lot more sound complexity and more dynamic range. It’s a backward song, in that the verses are pounding and the chorus is quiet, and you can clearly see the kathud at 0:56 which is one of the legendary drum and guitar collisions of all time as the first verse starts. Almost no clipping is seen, and you can see difference in the channels.
Now 1996. We see an album with the sticker 78:59, and we hear the word encumbered applied to music for one of the first times. Metallica proudly lays down 14 tracks very complex tracks with a much bluesier but still heavy sound, and then the label calls them in. The CD format maxed out holds roughly 80 minutes of classical music on a disk, and slightly less metal. We can’t fit this without “skipping” (i.e. compressing the master), you’ll have to cut a song. No f*$&ing way. In a compromise, “The Outlaw Torn” has its exit solo edited back by almost a minute to fit on the Load CD. There are huge differences between L and R on this track – listen to the growl vocal disharmony in your left ear when you have headphones on – and the bass is very pronounced. We’re now fully digital and notice the clipping going on, especially at 6:27. If my face becomes sincere, beware, but it’s just the beginning.
Now yesterday. In a complete reversal of history, Metallica leaks new music as digital tracks, so this has been out for a while. Just released on CD in the US yesterday, the Rick Rubin produced masterpiece “Hell and Back”. It’s a great song, probably one of my Metallica favorites lyrically (“tragic, heartless, and hateful” … thus the title of this post). I picked this because it compares to the first two examples in length with a quiet intro. Really only one thing to notice here: pretty much everything is clipped after 0:28, it’s hard to tell the difference between L and R, and it’s noticeably much louder than the other three tracks at the same volume setting.
Why is clipping bad? After all the Nyquist information theory is applied, and DSP does it’s magic, in the end it’s analog sound that comes out of analog speakers into analog ears. Clipping creates distortion, which come out as pops, thuds, and crunching that’s really not there in the recording, but the analog portions of the system can’t handle what’s going on. (Proof of that: some of the tracks from “Death Magnetic” have been remastered, and there’s a big difference.) Everything above the clip is essentially lost, forcing your brain to fill in which it tends to do until things get very messed up or it continues for a long time. There’s one really painful example at 5:02 in this song if you want to try it on headphones. Loud itself isn’t bad, but distorted is evil, and it’s distortion that blows up speakers and eardrums.
Yet, we have Rubin’s position. If most music is consumed digitally, via download, on a digital device with cruddy earbuds and no dynamic range anyway, and on top of that the user is in a high ambient environment and can’t go to full ear covering headphones, louder is better. You don’t need dynamic range if the player can’t reproduce it, and listeners today want something to fill their ears – quiet and subtle isn’t what they look for in music. What listeners want is loud. It’s the 2012 version of “gotta cut it down to 3:05″. It’s his job to produce music that sells, on the player of choice.
So, Neil Young sits down with Walt Mossberg and laments all this, invoking iGhost to help his case. Mossberg supported this:
Jobs in the past expressed surprise that “people traded quality, to the extent they had, for convenience or price.”
I’m not. We have lossless algorithms for sound, but they produce huge files. That’s a problem for not only device makers and users who want more on a device in less time to download, but companies like Pandora who stream and would need a lot faster infrastructure or risk putting their user experience on the line.
This may be one of those cases where user experience dictates “good enough”, and the premium of better doesn’t make a whole lot of sense. What do you think of the situation, Young’s comments, and the state of digital music from either a business or listener standpoint?
