The science of audio is mature, its beginnings dating back to the mid-1850s. In the same time span, myriads of new technologies have also matured, from internal combustion engines to human flight. Digital audio is younger, though still tracing its roots back some 75-plus years. That our experience of audio is rooted in human perception is where a difference lies in the acceptance of audio science as empirical fact, as opposed to, say, equating flight with metaphysics. Even many audio professionals are quick to dismiss the best science of the day if their personal preferences do not mesh with demonstrable scientific evidence.
A recent online discussion debated the merits, science of, and pure preferences for various audio delivery formats. It was a fascinating debate and inspired reflection on the progression of digital performance that I’ve personally experienced.
Once upon a time, I could reliably tell the difference between 16- and 20- bit recordings made from identical sources, mainly when the music wasn’t pumping, with reverb tails and fades, or on extremely dynamic material in an abnormal listening situation with a lot louder peaks than I’d normally listen to. 20- to 24-bit differences, I could only measure. I haven’t done those kind of blind tests in a while, but I’m guessing that any differences I could possibly hear with today’s top-notch converters and without pumping up low-volume material would require a top-flight monitoring system in a very quiet room and very open content.
I’ve been an advocate of high digital sampling rates, to a point. At one time, you needed 20-bits to get a linear 16- to 18-bits. Higher sampling rates pushed the bandwidth up to where analog anti-aliasing filters were not impacting the audible band as oversampling and digital domain anti-aliasing filters were not yet in refined, widespread use. 88.2 kHz, 96 kHz sampling or beyond may have value in production and within certain digital algorithms, but for final mixes, I’ll confess that I’ve not ever experienced, blind, a difference I could quantify for extremely high rates. I’m not sure that with today’s best converters that I could distinguish blind between 88.2 kHz and 44.1 kHz sampling. As for bit rate, 24-bit files and floating point math make all kinds of sense in production, but not necessarily in delivery.
I’ve had some amazing high res digital audio experiences with high sample rate material, but I’m willing to consider from a technical viewpoint that that could be as likely be because top-notch engineers were given the freedom to make the best recordings they could without pressure to hyper compress. I’ve also had some amazing listening experiences with CD-quality material where the loudness wars were not being fought.
The general public has embraced low-resolution, lossy digital audio primarily because it suits their lifestyle. Audio consumers could significantly improve their listening experience with a modicum of effort, facilitated by cheap digital memory and faster internet speeds. A marketing and education campaign could conceivably raise the bar of acceptable quality; audio really could use the same kind of campaign that video has benefited from. How many consumers learned that 1080P was “better” without having any clue what it meant? But they wanted it.
While degreed and professionally certified engineers in various scientific fields might be dismissive of audio professionals using the word “engineer” to describe their profession, the use of the term is applicable. Beyond the science and technology base implied by the word, there is also an undeniable aspect of artistry in audio engineering, defined by aural perception. Subjectively, what sounds good is good. The tools that help audio engineers realize an artist’s vision are the right tools for the job. That said, we can celebrate artistic brilliance without denying science by quantifying when our preferences cross the line from technical performance to emotional and visceral experience.