Thursday, July 10, 2008

Born To Compress

I wanted to follow up on the post I did on dynamic range compression in popular music. Actually, I plan to do a series of posts on the subject in which I compare recently released CDs to their LP counterparts. Today I compare the 2005 remastered reissue of Bruce Springsteen's Born To Run to a garden-variety 70s or 80s pressing of the 1975 LP.

Before I get into this, I want to say a few things upfront: I am not an expert on audio engineering and I don't claim to be. I've never set foot in a recording studio or even spoken to an audio engineer. I have absolutely no familiarity with the tools audio engineers use to record and master albums. I'm just a music fan with a set of ears, a computer, and a passionate interest in music. I don't claim there's anything scientific or definitive about what I'm doing. I'm sure there are valid questions that could be raised about my methodology (I have some myself), and I'm happy to answer them as best I can as they come up. Basically, I'm doing this to satisfy my own curiosity, and if I shed some light on the subject in a way that others find illuminating, great. So consider those caveats in reading my analysis, and see the comments section for technical details--how the vinyl to digital transfer was done, equipment used, etc.--if that sort of thing interests you.

I'm also not looking to generate a discussion about whether LPs or CDs sound better. I have my own feelings about that, and I'm sure you have yours. If you're looking for that kind of thing, there are many, many places on the internet where you can satisfy that desire. In my experience, debates about the relative merits of audio formats tend to go nowhere: they typically devolve into orgies of name-calling, accusations of deafness, and ultimately allegations of allegiance to Hitler. I don't want to go there. Whatever you happen to like is fine with me.

As I noted in my previous post, I suspect that Sony's remastering engineer added some extra dynamic compression to the 2005 reissue of Born to Run in order to bring the CD more in line with current industry conventions (i.e. to make it sound louder). After having ripped the title track from my LP copy of the album and comparing it to the CD version, I am further convinced of this.

I used Adobe Soundbooth to analyze the tracks this time. Unlike Fission, Soundbooth shows you both the left and right channel separately. It also shows a volume level reference in dBs on the right hand side of the diagram (you can click on the diagrams below to see them enlarged). Otherwise, what you see in Soundbooth is pretty similar to what you see in Fission. The closer the green lines are to the top and bottom of the chart, the louder the sound is at any given moment.

The most obvious difference between the two tracks is that, on average, the remastered CD is much louder. Excluding the fade out at the end, the remastered CD track ranges from around -6 db to just below the maximum limit of 0 db. But as you can see, quite a lot of the music is up very close to the 0 db limit. By contrast, the LP track ranges from around -12 dB to close to the limit of 0 dB. Also, relatively little of the the LP track goes all the way up to 0 dB. In fact, it only gets that loud for a very brief moment near the end of the song (right around 4:20). The average volume of the remastered version is probably around -2 dB, while the LP probably averages around -6 dB (I'm guessing here, anyone who knows how to calculate RMS in Soundbooth, let me know).

Bruce Springsteen - "Born To Run" CD (1975, remastered 2005)

Bruce Springsteen - "Born To Run" LP (1975)

Other than that the tracks actually look pretty similar. Loudness aside, in many respects they sound pretty similar to my ears too. I must say, I hear a certain muddiness in the LP version that has been at least partly cleaned up in the remastered version. The remaster sounds a bit "cleaner" and brighter to my ears. It also must be said that while there is very little evident dynamic range in the 2005 remaster, there is not a tremendous amount of dynamic range in the LP version either. Springsteen was going for a "Wall of Sound" presentation on this track and he succeeded. It's worth noting that this was a relatively compressed recording to begin with.

But there are a few moments on the LP version where the dynamic range is much greater (seen as brief "spikes" on the graph). I want to focus on a couple of those moments because these are the kind of contrasts that are inevitably lost when a mastering engineer chooses to push the average sound level as loud as is the case on the 2005 Born To Run remaster. The sounds made at those moments are still there on the CD, they just don't stand out as much with respect to the rest of the music as they do on the LP. Loudness has its price.

As anyone who has ever heard a performance of Tchaikovsky's 1812 Overture can tell you, sudden changes in dynamic range can have a tremendous impact on the way the human brain perceives music. When music suddenly gets much louder or softer it can startle and surprise us. Used properly sudden shifts in dynamics can create a shock or jolt that makes a piece of music more exciting, and can increase the listeners' engagement with and emotional reaction to the music. "Born To Run" is a textbook example of that in the realm of popular music recording.

The best example of this in "Born To Run" occurs right around 3:00 minutes into the song; the music builds to a dramatic crescendo, then briefly quites down. Next you hear the Boss count off "1, 2, 3, 4..." Bam! The band suddenly kicks in at full force again "Highway's jammed with broken heroes on a last chance power drive..." It's the song's single most dramatic moment, and fortunately the dynamic shift in that section is relatively well preserved on the CD remaster. Without it, you really don't have the same song.

But there are other similar, but more subtle, examples of this technique that get completely obscured by the remastering job. At right around :50 into the song you can see a very brief peak on the LP version that is buried in loudness on the CD. That's the exact moment Spingsteen sings "baby" on the first chorus of "Tramps like us, baby we were born to run." It makes a difference. Springsteen packs a lot of emotion into that one word, and hearing it at a higher volume relative to what's around it emphasizes that.

Another good example occurs almost at the end of the song, right around 4:20. That big peak you see on the LP version is the final cymbal crash before the fade out. The relative loudness of the cymbal crash puts a sonic exclamation point on everything that's come before it. Unfortunately, the CD remaster does not preserve that dynamic moment. The cymbal crash is still there, but it's no louder than anything else, so it doesn't have the same visceral impact as on the LP.

There are other moments like this on the LP that are mostly lost on the remastered CD. But what--if anything--do these differences mean in terms of listening experience? In a word "goosebumps." For me personally, the LP version, despite its relative murkiness, never failed to deliver goosebumps. Whether I was listening to it in my car, on headphones or on my home stereo, I consistently had a more visceral reaction to the LP track. I tapped my toe and pumped my fist more often when listening to the LP version. I just felt the music in my gut more. It's not like listening to the CD was painful. In fact, on an intellectual--if not visceral--level, I appreciated the added sonic clarity of the remastering job. With the CD it sounded like a slight layer of sonic murk had been scraped off the music, and I appreciated that. I wish that clean-up operation could have been combined with the superior dynamics of the LP version.

The experience of listening to the two versions was not the same, and invariably in ways that did not favor the remaster. The LP version of "Born To Run" is a one-of-a-kind, rock-and-roll thrill ride. It's the equivalent of a ride on Coney Island's Cyclone, while the remastered CD was more like a ride on one of those generic "Himalaya" style rides that travel from town to town with the carnie folk. It's true that the carnie rides go around and around really fast (and they're often really loud too), but they're never very exciting. A ride on the Cyclone though, that's something you never forget.


Pete Bilderback said...

For those who care, here are some technical details:

For the LP track, the turntable used was a Music Hall MMF-7 Turntable with a Grado Statement Sonata cartridge and Pro-Ject Speed Box ii. The phono preamp was a Grado PH-1 which directly fed an M-Audio Microtrack 24/96 recorder set at 16 bits/44.1 kHz (which is the same sampling bit size and frequency used on CDs).

Once I loaded the track from the M-Audio onto my computer, I imported the track into Adobe Soundbooth on MacBook Pro Computer. The M-Audio creates .wav format uncompressed digital files. I converted to the file to .aiff in Soundbooth, which really doesn't make any difference, but I wanted it to be the same type file as the CD track.

I did very little editing of the LP file in Soundbooth. I didn't do any click and pop or other noise removal/reduction. I did normalize the track, which raises the volume, but doesn't compress the signal (e.g. it raises the volume of everything equally without applying any limiting). According to Adobe: normalization adjusts "the highest peak of a waveform so it nearly reaches the digital maximum, 0 dBFS, thereby raising or lowering all other peaks accordingly. The Normalize feature in Soundbooth normalizes audio to 0.3 dBFS, avoiding clipping while ensuring optimal volume."

The CD file was an uncompressed .aiff file imported directly from the CD to my desktop then opened in Soundbooth. I did not do any editing on the CD track.

Next week I plan to do a comparison between Bob Dylan's Modern Times on CD and LP, using the same methodology. I plan to do some other comparisons in the future. Future posts on the topic will likely be a lot shorter than this one (but that's not a promise).

Charly said...

i forwarded your first post on this subject to a good friend who works A&R at sonylegacy in new york. he was impressed with your piece and said YES, he thinks that sales are affected by the loudness now part of many (most?) new releases. you are onto something real and important. thanks

Anonymous said...

"...and ultimately allegations of allegiance to Hitler. I don't want to go there. Whatever you happen to like is fine with me."

Sounds like you're saying it's just fine to swear allegiance to Hitler. I guess that kind of moral relativism is considered "cool" among the smugnoscenti, these days.

I would suggest, though, that this lack of a bright line between right and wrong is what leads audio engineers to believe it's ok to tamper with sound compression, if a certain segment of the populuation is "down with that."

Our actions have consequences, is all I have to say about it. You reap what you sow.

Pete Bilderback said...

Nice try Dan.

Adam Looker said...

You would have considerably more goosebumps from a 24 bit 96kHz recording for the original LP ;)

Pete Bilderback said...

Hi Adam. Thanks. Since doing this post I've changed my set up and now use an Edirol UA-1EX ADC set at 24/96 for vinyl rips. I agree that the higher resolution yields better results.

sam said...

I am really upset about the loss of dynamics. Being an avid music lover I simply cannot buy any remaster after the mid 90's because I am afraid that I am simply buying the inferior product. My favorite recordings are LP rips in 24/96 they seem to just be better quality recordings with all of the intended emotion and details. I think something should obviously be done but I feel like as an individual I have very little power in this.
Sorry for the long rant. The latest Beatles remaster is RIDICULOUS and has me shaken up.

I would love to get my hands on that 24/96 rip by the way!! ;)