The recordings are analog, but the medium is ink on paper.
This article describes the process used by Indiana University, and contains recordings from the nineteenth century.
It’s a shame that good articles like this 11-year-old piece have fallen into the Wayback-machine crypt. But they have.
A key conclusion reached is that not only does DSD suffer imperfections in accuracy, but also in precision. Ergo, it’s unlikely to produce the same waveform twice (due to intense noise modulation).
Of course, these technical imperfections may yield superior sound – especially when compared to 24/96.
I, for one, enjoy the DSD sound qualities (as do most folks that try them).
This video contains the most lucid instruction on digital audio theory that I have seen.
The exposition should benefit non-technical folks as much or more than techies, but beginners may want to start here.
In my piece, I was just a bit dismissive that capturing ultrasonic content could actually do harm. I’ve since eased my stance a bit.
You see, I actually downloaded Xiph’s test files and played them on my system. And low-level distortion products appeared. These were manifest as an occasional tic or pop. No wonder folks say that hires digital sounds analog! LOL.
Granted, my Goldmund amps have a rather large bandwidth (3 Mhz). But still.
Try the test yourself and let us know via the comments what you hear. The files are about a third of the way down in a section labeled “Intermod Tests.”
[Note that these occasional tics were down more than 40dB, and would likely be benign in a typical listening session.]
I’ve spent the last few evenings running through some circuit analysis using new web-based modeling software on Circuitlab.
So far, I love it! It’s fairly comprehensive, the GUI builder works well, and has nice graphical output. And it’s free.
Circuitlab is especially good news for Mac users, filling a void in application coverage.
The most “basic” use of the product is for passive crossover design, but active designs of all types are also doable.
So what’s stopping you?
Ensconced in my office at Untangle this afternoon, I was doing my daily regimen of high-tech reading when….I see an article on “24/192 downloads”appear on my favorite techie news service – Hacker News (HN). I read the xiph.org post - provocatively entitled “24/192 Music Downloads …and why they make no sense” - and made a comment or two in critique. My surprise was compounded when the post rose to the top of the HN chart…and stayed there.
Not only stayed there, but elicited hundreds of comments – from programmers! And the comments were permeated with subjectivism vs objectivism, room treatments, and of course pseudo-science. Just like the familiar audio forums! Could all of these geeks be closeted audiophiles? (If so, why aren’t they BAAS members?)
I don’t know the answer, but i will say that the article is well worth reading if you have an interest in how digital audio really works. Setting aside whether you buy its core premise, it is technically well-researched (includes links!), thorough, and lucid. With one or two exceptions, it’s also technically accurate. I think its major flaw is in vastly overstating the “high-resolution is actually bad” argument - unfortunate. The other glitches are quite minor, usually errors of omission rather than commission.
Enjoy it and, if you dare, comment on Hacker news!
1. The Xiph folks are champions of open-source media formats (e.g., Ogg Vorbis and FLAC). They generally know whereof they speak.
2. For the purpose of this article, I have assiduously avoided stating my opinions on the value of high sampling and/or bit-depth in reproduction. Rather, I have tried to focus on the technical merits of the Xiph article.
In this video, Ian Shepard of Production Advice in the UK compares a CD track to both its “raw” AAC coded form and a “Mastered for iTunes” rendition. I think that most of you will jump to the punchline early on, but the journey is interesting (for audio geeks).
On a related note, I wish that the audiophile industry would do more null testing to answer simple questions like: “does this cable sound different from that one?” It’s really not that difficult.
Audiophiles are taught that equalization (EQ) and compression are BAD. However, used judiciously, they are useful or even GOOD.
Joe Gilder’s “Home Studio Corner” web site is one of the best resources on the web for learning how good-sounding recordings are put together. No, Joe’s home studio is neither Skywalker nor Abbey Roads in sophistication. But it’s quite nice, and Joe is a natural and giving teacher.
The material on EQ is a good building-block for the room correction work that we’ll do. The compression video is also quite interesting, showing that the technique has uses beyond the “loudness wars.”
Few audiophiles share my passion for listening to test signals when evaluating equipment. That’s quite understandable.
But all would agree – from JA on down – that test signals are vital in measuring equipment and rooms. When measuring, the most common test signal in use is “pink noise.” But why?
Common answers include:
- It contains all frequency components
- It’s easy to use/generate
- It contains equal power per octave, and humans “hear” power; and
- It is natural
All of these are, of course, correct. But there’s a reason far closer to the audiophiles’ world: the pink noise spectral signature closely resembles that of music. It’s just that simple.
Not convinced? Compare the pink spectrum above to that of this Corinne May song that I chose at random. Except for the bottom few octaves, they share similar 10dB-per-decade falling slopes. This is not a coincidence.
The relatively weak bottom end (compared to pink noise) of this song is typical. Many recordings do not match the pink spectra down low. I believe this is a combination of music lacking sustained bass, EQ choices when mastering, and some loss of resolution of the FFT. But I don’t know for sure.
Lest you lose all hope for bass on CD’s and digital analysis thereof, check out Cowboy Junkies’ “Helpless” and Sting’s “Fields of Gold” (at bottom). These tracks are two of my bass references. And finally, we come to Max Richter’s “Blue Notebooks.” I know of no cut in my library with more bass extension -it’s nearly flat to 15hz!
You’ll get to hear a bit of pink noise at our upcoming room correction events.
The professional engineer in question has requested anonymity, but here’s what he had to say:
Studios are full of HF signals that alias back into the audio spectrum.
The guess about a TV monitor
somewhere in the mastering studio is valid indeed, as 15,750 * 4 -
44,100 = 18,900 Hz. It could even be a nearby TV transmitter.
Such “birdies” of constant frequency are relatively easy to trace, but
wideband HF is harder. For example, decent condenser mikes and
preamps go well beyond 100kHz at their output. Feeding them to
poorly-filtered ADCs (nearly all of them are!) can produce -40 to -60
dB of aliasing into the audio spectrum. As a side note, this is a
killer justification for ribbon mikes on “comb-spectrum”, high dynamic
range music such as authentic-instruments baroque.
Studio engineers can shrug and call HF birdies “digital sound”, yet to
me everything is analog: some within the audio band as intended, and
the rest just gets there unintentionally. Higher sample rate (192 or
176.4ksps for baroque, 88.2ksps for everything else) dramatically cuts
most of the latter.
See the comments of the previous post for more ideas.
Those of you who have attended events at my house know that I use this version of Sting’s “Fields of Gold” as one of my reference tracks (especially for bass).
I happened to be analyzing those bass peaks today, and noticed intermittent frequency spiking at 19khz (see image). To my knowledge, this behavior doesn’t occur naturally – certainly not in music.
Sure, FM uses a 19khz carrier tone to indicate the presence of stereo information. And Rich Pell has documented this and other high-frequency peaks on CD recordings. He cites speculation regarding the presence of video gear during the recording process. I think that there may be a tie-in to digital-audio tape, but there’s certainly no “smoking gun” in that regard.
My gut tells me that these peaks are simply an artifact of the digital filtering process. They are aliasing or some other frequency fold. But this too is speculation.
One thing that is not speculation is that they are prolific. I frequently (sorry) encounter them in my measurements. What’s worse is that they exist in some high-resolution recordings from famous and well-regarded sources. I don’t want to turn this article into an expose, so I’ll just leave it at that.
The good news is that I believe that these glitches are sonically benign for 99%+ of adult listeners. We simply cannot hear these frequencies, especially when they are narrow-band and 20dB-or-more down from the musical peaks. But they may drive kids crazy.
Physicist Glenn Elert has included a nice exposition on the mathematical basis of music in his online physics text (click here).
Glenn keeps the tech description at a “Popular Science” level for much of it, but descends to undergrad-college-level in parts. So there’s something for everyone. (nice illustrations too.)
If you think about the content, I believe you’ll emerge with a thing or two to enrich your listening – at home or in live venues.
PS The pic at right shows some anamolies I’ve detected on a popular DVD-A. I am shocked and amazed at what I find in some “hi rez” material. (At least this one goes above 22kHz – some don’t.)
Note that this ad hoc demonstration does not represent proof, let alone applicability to home playback environments. Also, extensive use of caching in player software would seem to negate this effect in sound systems.
However, perhaps this is why some austensibly-sensible audiophiles report better sonics with solid state drives (SSDs)?
Thanks to PeterT for pointing this out to me, from reporting in the San Francisco Chronicle.
We audiophiles are obsessed with frequency response (FR):
- Is that system “flat” to 20hHz?
- How low will these speakers go? I *must* have 20hz!
- There was a “BBC dip” in that Tannoy. No, I would call it a smile.
- Urgh. Those speakers are waaarrrm.
- And so on….
Even looking at the august Stereophile, keeper of the truth of measurements, we see JA focusing primarily on (for loudspeakers) impedance, FR, and time domain criteria.
What’s usually missing, often ignored, is the amplitude response (AR). How loudly will the system play distortion-free, and how softly? As any Lowther or horn fan would say, this dimension is also vital to producing the illusion of a live event.
Wait, you may say, everybody (including JA) publishes sensitivity data (dB/V for speakers, and dB gain for amps) and uses said data for buying decisions. True enough. But our present system of metrics makes it difficult to answer the original question regarding distortion-free amplitude envelope.
More fundamentally, before we even worry about what our own system can do, we might ask the question: how much is enough? How loudly (and softly) must my system play to reproduce “live levels”? (Assuming one cares. And BAAS members seem to have wildly disparate opinions on this point.)
While researching this issue in the context of my own system, I discovered an excellent survey article on the issue of audio dynamic range (click here). The source was a Greek website (unsurprising given the audiophile mania over there). I will warn you that the piece is long, winding, and – given the blazing red background – hard on the eyes. But it rewards the reader with a trove of cool data, from microphones to the digital chain and (eventually) out your speakers and into the room. Along the way, we investigate the noise floor of Davies Hall and Skywalker and more.
I’ll give you the gist. How much dynamic range do I need? More than you think. Can my present system do it? Almost certainly not. What can I do about it? Ummm, 24 bits, 500+ watts, multichannel….
Read the article. It’s a bit of work, but you’ll likely learn something new and useful.
Enjoy your Thanksgiving.
Yes, the title of this post is correct.
We audiophiles read a lot about how our components should measure, i.e., how to quantitatively specify system performance. You know the drill: 20-20Khz, 8 ohm, slew rate > 10, noise < 100dB, etc.
Sure, these answers can be found buried in textbooks and AES research papers. But few, if any, easily-accessible references are available.
Now, thanks to BAAS member Nyal Mellor, we have such a document. In fact, he and co-author Jeff Hedback wrote it! And they graciously donated a copy to BAAS for download.
You will need to use FTP to get it. See the new “Downloads” button above for instructions. It’s easy!
I wrote earlier in this blog about how digital audio signals must ultimately be represented by analog signals – particularly when traversing a cable.
This article in EE Times, while geeky, presents more evidence about the slippery problems of digital audio – this time from the perspective of USB.
Don’t think USB cables matter? Read on – if you dare!
The geometry of Nordost Odin. The circuit-in-a-cable approach of MIT. The static charge on Synergistic shielding. All examples of how audio cables have progressed beyond the twisted pair.
But these are childlike in sophistication compared to the Thunderbolt cable, as featured in new Apple Macintosh products.
T-bolt, which some suspect to be the ‘next big thing’ in computer-audio connectivity, houses a dozen I/C’s in each cable!
Check out the details in this iFixit blog piece.
Fascinating stuff (for a geek).
This just in from member Bob S.:
Click here for Ear Training software from Harmon Intl. Really cool, and it's free!Some bugs are discussed here.Finally, Figure 3 (here) shows the poor performance of a group of audio reviewers (how chosen?) compared to listeners trained by Harmon's method.Bob S.
This week, Intel and Apple announced the first incarnation of Light Peak – on the latest iteration of Mac laptops. Apple calls it Thunderbolt.
Cheesy name aside, the 10 gbps performance is staggering. It’s more than an order of magnitude faster than Firewire 800. Want to move a Blueray movie? Less than a minute. USB 2.0 and 3.0 just got leap-frogged, big-time.
For those of us with terabyte+ music libraries, moving these and backing them up just got easier. This interface has 3x the transfer rate of a high-performance SATA drive!
Is Apple positioning this as the successor to firewire? Probably. Will it lead to better fidelity in music server systems? Who knows.
But I want one…
I just wrote a lengthy post about digital crossovers and EQ over on the Computer Audiophile forums (click here to read it).
Bottom line, while not for everybody, DSP technology can revolutionize the way you interact with your system.
I also make some specific product recommendations that I thought would be inappropriate for BAASnotes (being as how we’re Switzerland and all….
- March 2014
- February 2014
- January 2014
- December 2013
- November 2013
- September 2013
- August 2013
- July 2013
- June 2013
- May 2013
- April 2013
- March 2013
- February 2013
- January 2013
- December 2012
- November 2012
- October 2012
- September 2012
- August 2012
- June 2012
- May 2012
- April 2012
- March 2012
- February 2012
- January 2012
- December 2011
- November 2011
- October 2011
- September 2011
- August 2011
- July 2011
- June 2011
- May 2011
- April 2011
- March 2011
- February 2011
- January 2011
- December 2010
- November 2010
- October 2010
- September 2010
- August 2010
- July 2010
- June 2010
- May 2010
- April 2010
- March 2010
- February 2010
- January 2010
- December 2009
- November 2009
- October 2009
- September 2009
- July 2009
- June 2009
- May 2009
- April 2009
- March 2009
- February 2009
- January 2009
- November 2008
- October 2008
- September 2008
- August 2008
- July 2008
- June 2008
- May 2008
- April 2008
- March 2008
- February 2008