Thursday, 5 February 2015

MUSINGS: The ongoing Vinyl vs. Digital debate...

The always assertive Batman. But assertiveness isn't necessarily correct...
(Just a shortish post this week... I've covered much of this in the past.)

As I mentioned in the comments section of the previous post on HRA, there have been discussions recently again around the sound quality of vinyl compared to CD. I assume it started from this article "Why CDs May Actually Sound Better Than Vinyl" from the L.A. Weekly. Overall, I think it's a good article! Some excellent quotes from veterans in the audio engineering business like Bob Clearmountain and Bob Ludwig. We even have the pleasure of an interview with James T. Russell - the "father" of digital optical media. Folks... These people know what they're talking about, so do not take their comments lightly.

Despite the fact that there's nothing ground breaking in that article, it's no surprise that the vinylphiles are up in a tizzy. Michael Fremer has now dragged out an old 1996 article about "Does Vinyl Have Wider Dynamic Range Than CDs?" (the PDF in the article). Let's talk about this...



Scroll down to the comment by "comfortablynick" in that blog post for a good discussion about some highly questionable comments made by the author around his beliefs about digital audio. The lack of consideration for the importance of dithering to 16-bits is a particularly egregious oversight. "Phuzzyday"'s comment about the author referring to "41kHz" as the CD sampling rate in the final page of the article also does not help the author's credibility.

Even with all these concerns, I want to lodge one objection to his basic "thesis". It's his definition of dynamic range for the purpose of this article from which everything else builds:
"For CD players, dynamic range is essentially the ratio of the loudest possible output signal to the quantization noise, i.e., the noise corresponding to the round-off error of the least significant bit. For the 16-bit standard CD format, this is 96dB dynamic range. 
To parallel the previous CD definition, I define the dynamic range of a phono cartridge as the ratio of the loudest sound to the background noise referenced to the preamp output terminals..."

As mentioned, the first paragraph is inaccurate in that it doesn't take into consideration the effect of dither on perception of a low level signal, allowing us to effectively hear below the 96dB quantization noise floor. With 16-bits, dithering will allow us to perceive down to about 110dB with "flat" dithering and even more with noise shaped dithering in the frequencies that humans are most sensitive to (see this page for an example using 8-bit audio to demonstrate the effect). Dithering has been standard practice since the dawn of digital media. Using the definition in the second paragraph, the author then produces calculations based on ideal values predicted by the electronics and mechanical velocities to estimate the capability of the phono cartridge-preamp system; coming up with up to 110dB for the systems analyzed. The problem obviously is that this level of performance is not what we experience when we put on a piece of vinyl for a spin!

Where in these calculations do we take into account the real world? Remember, vinyl doesn't have "infinite resolution". The LP itself and the playback system are imperfect physical object. Surface noise? Pressing inaccuracies? How about geometry issues with the tonearm and cartridge setup? Groove wear over time? Stylus wear? Noise picked up in the low voltage phono wires? Imperfect dampening? I'm sure there are other issues I've neglected.

Of course these issues will also result in other (IMO more bothersome) distortions and not just dynamic range limitations. For example, the temporal distortions from wow and flutter compared to the accuracy of digital playback (as I've said before, nobody should be concerned about DAC jitter if one can tolerate the inaccuracies from a turntable!).

In comparison, for digital, the data can be transmitted with 100% accuracy quite routinely. Already excellent with spinning polycarbonate CDs using laser light to read the information, and much more so with hard drives, or SSDs these days. The media itself (basically just the file with computer audio now) is "perfect" compared to the limitations of the physical analog vinyl. Furthermore, the conversion of the digital information to analogue electrical signal in the DAC is completely free from mechanical issues. Therefore the real world output characteristics like dynamic range is also much closer to the ideal (and more predictable compared to the variability of LPs from a store). These days, 16-bit dynamic range is quite easily surpassed with decent 24-bit DACs.

It's interesting that the author acknowledges noise in LP playback (p. 24) but he doesn't seem to estimate the (significant) contribution or incorporate this into his calculations. Needless to say if he did, the numbers would not be impressive. Probably just ending up as the "widespread notion that phono dynamic range is only 60-70dB" (p.17). Notice the sleight of hand here in that the author is only calculating the cartridge and preamp's theoretical dynamic range but the title of the article and implication from Fremer is that it refers to the whole phonograph playback system so as to have any meaningful comparison with typical CD playback.

In practice, realize that 60-70dB of dynamic range isn't that bad; that's about 11-bits of dynamic range in the digital world... In fact, take a typical 16-bit digital file and knock off the last 5 bit of resolution. It doesn't sound so bad still unless you're listening to stuff with lots of dynamic range and you need to up the volume - like classical music perhaps. It's odd that he seems to blame the RIAA; "lack of quality control is abetted by the RIAA record industry standard permitting surface noise of 55dB below a 1kHz sinusoid recorded at 7cm/s". Hmmm, again, if this is indeed the issue that limits the true dynamic range at playback, why isn't Mr. Bauman spending time to show us exactly how low surface noise can be and if state-of-the-art material sciences in vinyl is able to embed a signal anywhere close to his idealized calculations of dynamic range and show us how that would compare with digital!?

Which brings me to my last point. The classical music people I know are interested in CDs and SACDs (maybe DVD-A and Blu-Rays), not vinyl. Other than a few collectible classical albums usually on display, the used vinyl stores I peruse typically sell the rest at $3.00 or less. In fact, one of the stores has a massive rear portion with probably thousands of classical LPs that appear to go untouched; the stock doesn't seem to move and I rarely see people rummaging around back there compared to the rest of the rock/pop/jazz sections up front. Classical music demands silent noise floor for the quiet portions and dynamic range is truly an asset. I have no problem with the higher noise floor from vinyl with most rock and pop since it can be masked by the higher average volume, but classical music demands higher quality vinyl; it certainly makes sense for classical music lovers to go digital primarily. And as far as I can tell, this is exactly what has happened given digital's superior fidelity. As you can see in the James Russell interview with L.A. Weekly, he developed optical media because he felt LPs were not good enough for the resolution demands of classical music. (Furthermore, beyond sound quality, maybe convenience is even more important in classical music to avoid flipping sides and changing disks while trying to enjoy a full symphony.)

To end off, like others, I do enjoy my collection of LPs. I'm up to >300 albums now (out of that only a handful of classical). They live side-by-side with my digital music harmoniously in the sound room. No problem switching between LPs, digital files, multichannel music in an evening of listening. They each have their pros and cons. LPs can sound fantastic when the vinyl is in good shape, and the artwork can't be beat. The cleaning, cataloging, playback ritual provides a physicality that blesses and satisfies the collector's soul (or conversely a curse to the obsessive-compulsive hoarder). Personally, they bring back old memories of a time in my youth (70's & 80's) when I could not afford to own this stuff (but much of the popular albums I like are now cheap!). But it is old technology, superseded in all kinds of ways by the potential fidelity* offered with CD and now digital files. Let us not kid ourselves about the inherent limitations of the vinyl format when it comes to reproduction accuracy/fidelity compared to the original "studio master". I hope we can be graceful with the acceptance of this fact rather than with denial and anger.

One final thing... Cassette tapes making a come back? Sure! I love the scenes in Guardians Of The Galaxy. Just don't call it "high fidelity" or better than MP3, OK? :-)

Exhibit A (from a few years back - for kicks, check out the YouTube comments on this!):





* Take note that I said "potential fidelity" here. There are many LPs that clearly sound superior to the CD counterpart. This happens all the time with better mastered (eg. more dynamic) LPs compared to terribly dynamically compressed CDs. The "vinyl rips" to digital files sound fantastic with these LPs and demonstrate just how well digital technology is capable of capturing what vinyl has to offer. But if we were doing a direct comparison with exactly the same mastering on both LP and CD (using the best LP cutting system and best ADC), I'd go with the science and say that the CD digital format would be able to reproduce the frequencies up to 22kHz with greater precision both in terms of dynamic range and time accuracy; it would sound more like the original master. LPs can reproduce >22kHz frequencies if the source contains ultrasonic material but evidence for needing ultrasonic frequencies to improve audio fidelity remains elusive (see the HRA post last week and the Oohashi stuff).

----------------------------

Alright folks... Lots of "real" work to get done now. Have a wonderful weekend and week ahead.  See you in the next instalment (I might be away a couple of weeks). Enjoy the music - in whatever format you darn well choose!

3 comments:

  1. I am in no way aiming this comment at the author of the blog BTW but I think "better" is a term worth exploring as it seems to be a matter of perspective.
    We have cases of respected music produces claiming that high-res recordings are "better" and they are probably right (from one perspective). Since they were at the bridge mixing the album themselves, they would be the final arbiter of what the album sound(ed) like (Yeah, OK, through their speakers and/or headphones which WILL alter the sound more than anything in the audio chain!). The high-res recording is "better" in the respect that it replicates as closely as possible the sound that they were getting at the time of recording or at least what was on the master. The master would then be disseminated through various media - records, tapes, mini discs, CD's, DVD's and now digital. Each format does things to the sound that the consumer hears. Records have a low dynamic range and lots of crosstalk, tape had hiss and distortion at high dB, redbook imposed a brickwall at 20khz and so forth. Back when the recording was released, the consumer might have bought the album on vinyl. That was their point of reference. It would have sounded great and the producer would have most likely mixed the album with the media in mind. The fact is, I think high res sounds great and I think vinyl sounds great. One recreates the original in the most accurate way possible and the other makes the same album sound very different but in a very pleasant way. Is one superior to the other? On purely statistical grounds yes. If you enjoy listening to high-res then do so (I do out of purely practical reasons). If you hate the coldness of digital, go any buy vinyl - and enjoy it. It would be great if we could stop getting bogged down in arguments about "better" and accept each format for what it is and stop arguing!

    ReplyDelete
  2. May I ask what turntable gear you use? Its seems a lot of the snake oil involved with digital is a throwback to the things which did matter for vinyl but do not for digital (such as disc cleaning, isolation) and such. I cannot et clarity on whatt he sensible objective purchases are for vinyl. I use a Benchmark DAC and Ncore amps, so value good measurements and a lack of hot air! Thanks in advance! :)

    ReplyDelete
  3. The theoretical superiority of digital is simply indisputable fact. The "digital remastering" craze of the '90s however threw those advantages out the door. Take for instance Led Zeppelin's original masters which were not particularly great to begin with and compress them further and you can make them downright awful. I remember how excited I was to rip the seal off the remastered LZ III and to pop it in in my (then shiny new) Rotel only to be terribly disappointed after the first 3 tracks. Ended up buried in storage somewhere, never to be played again...

    In practical terms this digital superiority failed to materialize until around 6-7 years ago or so but even today new releases tend to have an analog component, the digital being reserved for streaming/ download/ mobile use thus ending again compressed for use in relative noisy environments. Perhaps a visit here would be more ilustrative http://dr.loudness-war.info/

    Finally, the amount of new LP releases and re-releases speaks volumes for the viability of a medium that was supposed to be dead 25 ago. I for one enjoy this so called "revival". My 15 year old daughter as well! Adele, Tame Impala... you name it, she spins them all, along brand spanking new Pink Floyd or LZ LPs.

    ReplyDelete