Recently, I was alerted to an article published in HiFi News & Record Review, a British audiophile publication that claims to be "the longest serving and most prestigious hi-fi and music magazine in the world". It was written by Keith Howard, an acquaintance of mine from some years back, and entitled, "HD download debacle". The subtext reads, "High sample-rate music downloads are not all they seem". I couldn't agree more and was quite pleased that the publisher of a major magazine on audio has the wherewithal to take the high ground on this issue. My compliments to editor Paul Miller and HFN. Everything that Keith discovered during his investigation maps perfectly with my own research and reporting. The world of HD digital music retailing is not everything that we would like to expect.
I'd like to share a few of the items that Keith included in his report. I think readers will find this very illuminating. Here's his opening paragraph: "When audiophiles buy a hi-res music download, most do so on trust. If they've paid a premium for a 24-bit/88.2, 24/96, 24/176.4 of 24/192 download, they reasonably expect that the enhanced bandwidth offered by the higher sampling rate will be fully exploited, inasmuch as the source material allows. But our investigations show that this trust is sometimes misplaced, and those price premiums are being asked for audio files in which the signal bandwidth has been curtailed."
First, he points out that this is not a new situation. At the introduction of high-resolution or high definition audio formats back in 2000, Paul Miller published an article exposing many of the early DVD-Audio productions as lacking substantial improvements over CDs. The SACD and DVD-Audio formats were specially designed to, "demonstrate the audible superiority of 96 kHz/192 kHz recordings over CD's 44.1/16-bit format [but, in fact] actually sounded worse." My contention has always been that a standard definition recording from the past placed in a container that exceeds its fidelity standards remains a standard definition recording. We might be getting the best possible rendition of that older track but it is not the same thing as having a new recording done with live musicians at 96 kHz/24-bits. And it shouldn't be marketed as such.
Keith's first example of "ham-fisted" upsampling came from High Definition Tape Transfers, which to me is an oxymoron of the most blatant type. Every analog tape recording is standard definition (limited dynamic range and frequency response) thus transferring it to an HD bucket is pointless…unless the company juices the frequency response somehow. There are a couple of very informative graphs showing the "butterfly" effect of this sort of audio foolery. HDTT remade the files and sent them to all of their customers that had purchased the version on steroids.
The next part of the discussion in the article focuses on HDtracks.com, the company headed by David and Norman Chesky. Keith writes, "has never, to my knowledge, released anything so crass (as the HDTT folks) but is has sold, and at the time of writing continues to sell, files which do not have as wide a bandwidth as you might reasonably expect from their sampling rate."
He continues by pointing out, "as an example that's been on sale for a long time is the 24/96 download Peter Frampton's Frampton Comes Alive, the spectrum of which clearly shows the presence of steep low-pass filter just above 20 kHz." The track, he concludes, "this track has been upsampled." To be fair, Mike Lawson of HDtracks did re-label the Frampton title as 48/24 and it is as good as you will ever get from an analog sourced original. My argument is that it should be labeled accurately from the outset. Anything that goes back to the days of analog tape shouldn't be "upsampled" and sold for a premium price.
In a subsequent paragraph, Keith goes on, "HDtracks removed John Coltrane's Lush Life when this was exposed as being filtered. The spectrum of "Like Someone in Love" appears to have been low-pass filtered twice, probably indicating that it was upsampled from 44.1 kHz to 96 kHz. In contrast, the Frampton download remains available at the time of writing this and, moreover is not an isolated case - in fact it has some notable company among HDtracks' classical titles." He downloaded a couple of classical tracks that are available on the HDtracks site and found that they too, were subjected to "steep low-pass filtering just above 20 kHz."
And it gets worse when you think that customers can spend an additional $10 for the 176.4 kHz versions. Keith's conclusion, "the $10 premium for the 176.4 version buys you, effectively, nothing."
The article also targets Linn's high resolution downloads as suffering from the same manipulations. Linn promises to pay closer attention to the quality of their source, the rigor of their procedures and to do a spectral analysis of all new content. Why wasn't this done previously?
The end of the article doesn't instill a lot of confidence in the press and websites that report on our industry, "Unfortunately the hi-fi press - which ought to be taking a leading role - has mostly sat on its hands: hi-res recordings are routinely reviewed without any attempt to confirm their provenance. Web sites that review hi-res recordings are arguably even worse since their coverage typically outstrips that of the hi-fi magazine but their reviews again include no objective assessment of the signal bandwidth supplied. Online audio forums fill the gap to some extent, but aren't to be relied on, in this matter or any other. For instance, in an Audio Circle forum discussing HDtracks' Rolling Stones downloads, ted_b, described as a Facilitator, wrote, 'Spectrum analysis shows lots of energy way above 30 k for these Stones 176.4 k rips, and not just noise-shaping' - which clearly flies in the face of our own results".
I believe that it's time for digital music retailers, high definition record companies and the press (both printed and online) to adopt an open and honest approach to high definition music recordings. The more information that consumers have the better it will be for everyone…the high-end segment of the business will improve and music lovers will know what is possible with real high definition tracks.