Audio 2019

I like to say a few words, about once a year, about my now somewhat dormant audio hobby. “Audio” means the machines that make music. It is a way to play with machines, which men like to do. And also, it is a way to listen to music, with more involvement and satisfaction than you might get from compressed streaming downloads over Bluetooth to your car speakers, or cheapie earbuds plugged into your phone.

Last year, I talked a bit about my headphone setup. Headphones have become quite popular recently, as they have many advantages. You can get a very nice headphone setup for less than $2000, while this barely accomplishes anything if you have speakers. It is compact, doesn’t mess with your decor, and doesn’t bother the neighbors.

February 19, 2018: Audio 2018

Nothing much has changed this year, at least in the broad picture. I did replace my previous first-generation Metrum Octave DAC with the newer Amethyst DAC, also from Metrum. It is basically two generations later, and has many upgrades compared to the original. I did this for a few reasons: 1) integrated USB input. No separate USB/SPDIF interface. 2) upgraded internals including the Transient DAC module, and FPGA processing. 3) integrated headphone amplifier. 4) The ability to handle 24/192 files, the new standard for very high resolution, instead of the 24/176 max of the Octave. What I discovered is: the sound of the Amethyst was noticeably different from the Octave, but I was hard-pressed to say that one was better than the other. Both were very good. The newer one is “supposedly” better, in terms of specs and many engineering advantages, but I didn’t hear a major improvement. But, I thought I’d keep it anyway, because I might, in time, grow to appreciate the advantages of the newer, and eventually find, in the future, that if I tried the old one again its disadvantages would be more obvious. Second, the included headphone amplifier is quite bad, and basically unusable as far as I am concerned. I will keep my existing amplifier, the Schiit Asgard II, which I was earlier hoping to get rid of.

If you recall from before, I am using the Intact Audio 600-ohm autoformer volume controls to connect directly between DAC outputs and headphones, with no amplifier. This is very good. Very, very good. Since the Metrum units (old and new) also don’t have any output stage, but go direct from chips to output, we thus go direct from chips to headphones with no intervening active stages. Considering how much damage was done by a simple current buffer–the included headphone amplifier–I regard this as an excellent solution, if you can get away with it. Comparing the direct outputs with the same DAC through the Schiit amplifier, I do not find any disadvantages (such as weak dynamics) from the direct approach, at the same volume level. Some kinds of music might benefit from higher volumes, however, which is when the Schiit comes into use. But this is rare. With my relatively high-efficiency Sennheiser HD600 and Audeze LCD-2 headphones, I seldom want for more amplification. The Schiit, which is very inexpensive, seems to work quite well, and I don’t find much degradation compared to the direct outputs.

The last little discovery is that 24/96 over Audirvana does indeed sound quite a bit better than 16/44 “redbook” (regular CD quality). “24/96” means 24 bits per “word,” or sample, and 96,000 samples per second. It might sound like 24 is only 50% more than 16, but what it means practically is that eight more bits allows 256 times more subdivisions per sample. Since 24 is 50% more than 16, and 96 is about 2×44, basically a 24/96 file has about three times more data than a 16/44 file. For comparison, 16/44 “redbook” corresponds to about 1440kbps of data transfer, or about 700kbps in lossless compression. The typical streaming service (Amazon Prime Music) runs at around 200kps, so Redbook itself has about 3-4 times more data resolution than typical streaming solutions. I was previously of the opinion that 16/44 actually sounded pretty good, if the recording itself was well made, and that the additional advantages of 24/96 might add a little detail but didn’t really matter too much for what is important to me, which is a kind of sonorousness, what the late Harvey “Gizmo” Rosenberg called “King Tone.” I like gear that “sings.” To use a little bit poetical language, you want to gage how something feels, not how it sounds. Of course it “feels” a certain way because of how it sounds, but you can’t really perceive it consciously with your ears. Since musical enjoyment is our goal here, you want something that feels musically enjoyable, not something that sounds accurate. Of course it may also “sound accurate,” but there is a lot of bad-sounding gear that “sounds accurate.” In a similar way, you can’t judge wine because it is dry or sweet, or anything else that one can rationally analyze, among better-quality wines. The subtleties of flavor result in wine “feeling” different (wine fans don’t use these words but you know what I mean). It produces a different artistic and emotional response. It is a little like a Stradivarius compared to a common violin. This “sonorousness” I find often comes from electronic topology (in my case, eliminating electronics altogether). In the case of audio gear, instead of wine, we are really one level removed. The way the music “feels” actually depends on the performance. But, we want that “feeling” to be retained in the recording and reproduction.

So, I was slow to compare 16/44 and 24/96 side-by-side. I had some 24/96 recordings that sounded very good, but I assumed that they would also sound good on 16/44, because they were engineered and recorded well. I also had some 16/44 recordings that sounded very good, and I assumed that they would also sound very good on 24/96, but not a whole lot better, and not in the ways that I value, which is emotional and artistic effect. I finally got around to actually comparing the same recording side-by-side, which in this case was Absence by Melody Gardot. On my system as described, the 24/96 version sounded considerably better, and better in the ways that are important to me: as a kind of artistic and emotional expression, or “tone.” The 16/44 sounded “processed.” This was not hard to notice. I could tell in the first five seconds. And, once I came to recognize the “16/44 sound,” this peculiar “processed” tone, I noticed it in many of my other 16/44 recordings, which I previously thought sounded pretty good, but now I want to hear them in 24/96. So, I guess I will be spending more money at in the future. Music today is commonly recorded in 24/96, so we are in a sense listening to the original “studio master.” This result, however, is dependent on the electronics you are using. If I was using the headphone amp provided with the Amethyst for example, I probably wouldn’t notice much difference. The inferiority of the amplifier would overwhelm the subtleties of the recordings.

Most all pop recordings these days have way too much processing and dynamic compression, which ruins the sound quality. So, all of these considerations apply mostly to properly-recorded music, which these days means classical and some (but not all) jazz. Unfortunately, engineers feel a responsibility to mutilate their work these days, and artists go along with this. There are a few exceptions, with Neil Young a good example.

There is a lot of hype about statement-class headphones these days, costing more than $1000. I would recommend first trying some headphones that are premium quality, but not in this big-dollar class. A headphone is a pretty simple thing, and it should not be all that difficult to make a good one. There are a lot of classic designs in the $200-$400 range, from big makers that have designs that they have refined over decades. Besides the Sennheiser HD600 that I am using, I would include the AKG 701, Beyerdynamic DT-880, and the Grado SR325e. If later you try some big-dollar headphones and don’t find them to be much better, don’t be surprised. It’s a little secret among car fans that the best Corvettes are actually better cars than the Ferraris or Bugattis. And if you do find them to be much better, then you will know that you are indeed getting your money’s worth.

Those are my suggestions for to-the-limit headphone listening that doesn’t cost much. I don’t think you could do much better at any price. (I would try Metrum’s more ambitious DACs.) You could definitely spend more and get less, and almost everybody does. The secret sauce is the 600-ohm autoformers from Dave Slagle. But, this is a DIY solution. You have to wire it up yourself. There is this product, which looks similar although I am not sure of that:

“Line level headphone adapter, transformer based gain setting: $150”

I am sure Dave would be willing to wire some up himself to order, but it will cost you.