Is high-quality streamed audio an oxymoron?

Article By : Brian Dipert

Will cost-effective upgrades to computer-based audio playback hardware deliver sufficient results to justify the prices paid for them, particularly with lossy-compressed music sources?

Lossy compression has, if not already obvious to you, overwhelmingly captured the multimedia format distribution crown, not only for audio (the subject of this particular blog post) but also for still and video images. And, given the choice between low bitrate and high quality audio, consumers have overwhelmingly voted in favor the former option … although in saying this, I’m admittedly being a bit over-simplistic. After all, newer generations of lossy compression algorithms (AAC and WMA, for example) tend to notably outperform their predecessors (such as the ubiquitous MP3) at various quality metrics (high frequency response, SNR, distortion, etc.), especially at low bit rates. And, in spite of storage capacity and download/streamed payload concerns to the contrary, today’s music services generally deliver the tunes at 192-320 Kbps rates, versus 128 Kbps (or lower) with service precursors of times past.

That all said, it’s still lossy compression, at which discerning audiophiles snort with derision (then again, they turn up their noses at compression-free Red Book Audio CDs, too). Are they right? Or, at least for those with less-than-pristine sonic discernment (not to mention less-than-infinite sized wallets), are there cost-effective upgrades to elementary hardware that will deliver ample results? In a recent post, I discussed one possible improvement path: adding a subwoofer to extend the playback frequency response. In that same post, I alluded to another planned Echo upgrade (and blog post covering it) to come: replacing the mono primary playback speaker with a matched stereo pair. And this time, I’ll talk about another case study of the concept, this time focused on computer-based audio and looking not only at the transducer (headphones in this case) endpoint of the playback chain but also at the intermediary digital-to-audio conversion and amplification stages.

Showcased throughout will be products I originally purchased at Drop, which many of you might know from its former Massdrop moniker. To be clear, I’m solely a satisfied customer of the company, which partners with manufacturers both to supply bulk “builds” of products at cost-effective prices in exchange for guaranteed customer numbers, and (in some cases) to develop custom variants of those products. Drop has no idea I’m writing this piece, or even that this particular customer is also a technology writer; I’ve paid for all of my purchases completely out of pocket, only occasionally supplemented by “you haven’t bought from us in a while” discount coupons they sent me.

The music library I leveraged for my auditions was extremely diverse in genre (everything from classical to heavy metal, along with numerous in-between styles) and came from two primary sources: downloads from the Apple iTunes (now Music) Store (16-bit per-channel, 44.1 kHz (PDF)) and streams from Amazon Music Unlimited (unknown per-channel bit depth and sampling rate), both in 256 Kbps AAC. In the latter case, to ensure high bitrate, I wasn’t able to stream directly from a web browser; instead, I downloaded the dedicated Amazon Music app for PC and Mac and then dove into the settings and changed the default “Auto” audio quality variable to “High”:

The Amazon Music apps for Android and iOS are similarly configurable, by the way:

I intentionally focused on today’s most common bitrates, in the process avoiding lower bitrate options whose increased amount and variety of lossy compression artifacts might swamp any equipment-induced differences. For similar reasons, I avoided even higher claimed quality audio sources such as the content available for purchase from HDtracks … since this study focuses on audio for the masses, I felt that niche services like this might unduly magnify any hardware variances.

The quality potential inherent in any audio source won’t be discernable if you listen to it over cheap (i.e. limited and uneven frequency response, high noise and distortion levels, etc.) headphones, of course. To simulate a range of usage environments and price ranges, I went with three different sets of headphones; the $29.99 low-impedance (60 ohm average across the measured frequency range, to be precise) Massdrop x Koss Porta Pro X:

$149.99 150 ohm open-back Massdrop x Sennheiser HD 58X Jubilee (example review here):

and $199.99 300 ohm (also open-back) Massdrop x Sennheiser HD 6XX (again, here’s an example review):

Why the range of impedances? The subject is controversial, to say the least; here’s an in-depth treatise on the topic (which references other writeups I also read in doing my research). Here’s a concise summary: all other factors being equal, a higher-impedance set of headphones will generally give you higher quality audio reproduction, for the reasons cited here:

The impedance of a headphone is largely determined by the driver’s voice coil, and for Beyerdynamic’s high-impedance models the voice coil’s wire is super-thin, just 0.018mm, half the thickness of a human hair. Beyerdynamic’s Senior Product Manager Gunter Weidemann explained … The thinner wires have more windings (layers of wire) on the voice-coil than the lower-impedance Beyerdynamic headphones, which have thicker and heavier, easier-to-manufacture voice coils. The lower moving mass of the 250- and 600-ohm headphones’ voice coils is lighter than the 32-ohm models, and the lower mass is part of the reason high-impedance headphones sound better. The smaller diameter of the 600-ohm voice coil wires allows the wires to fit tighter, so there’s less air between the windings, and that makes the electromagnetic field of the voice coil stronger. All of that reduces distortion for the high-impedance versions compared with the low-impedance headphones.

However, higher impedance headphones require more amplification in order to be able to successfully drive them to a given volume level, which is a particular challenge for low-voltage battery-powered portable audio devices (for which long battery life is also a key consideration). This explains why low impedance headphone models also exist and, in fact, now dominate the market in terms of volume shipments. Portable device compatibility was, for example, a key factor in the decision to develop the Massdrop x Sennheiser HD 58X Jubilee.

For completeness, I should also point out that accurate discernment of audio quality differences is also heavily dependent on the potentially degrading effects of competing ambient environment sound levels, especially when open-back headphone models are in use, which is why I did all of my auditions in a quiet room.

In the next post of this particular blog series, I’ll discuss the external A/D converter (ADC)-plus-headphone amplifier combo I tested against the equivalents already built into my laptop, and share the results of the comparisons I auditioned. Until then, “sound” off in the comments, readers!

Brian Dipert is Editor-in-Chief of the Embedded Vision Alliance, and a Senior Analyst at BDTI and Editor-in-Chief of InsideDSP, the company’s online newsletter.

Leave a comment