High Resolution Audio (HRA)

High Resolution Audio claims to provide near-live performance quality (as opposed to CDs and internet streaming) but technically, HRA simply refers to lossless digital formats with more than 16-bit depth and more than 44.1 kHz sample rates.

Claims for HRA (24-bit / 96 kHz and higher) digital files include ...
  • Myth #1: HRA will directly produce an enhanced listening experience simply because of 24/96
  • Myth #2: HRA versions of existing recordings will always provide higher quality playback
Yet, double-blind testing results suggest ...
  • Reality #1: 16/44.1 FLAC is indistinguishable from 24/96 FLAC
  • Reality #2: 'Mastered-For-Loudness' 16 or 24 bit recordings is the root cause of degraded recordings
One, of several, double-blind tests has already confirmed that a wide range of listeners are statistically unable to detect the difference between 16/44.1 and 24/96 resolution -- when that difference is ONLY the bit depth and sample rate. Here are the links:
  • http://archimago.blogspot.ca/2014/06/24-bit-vs-16-bit-audio-test-part-i.html
  • http://archimago.blogspot.ca/2014/06/24-bit-vs-16-bit-audio-test-part-ii.html
  • http://archimago.blogspot.ca/2014/06/24-bit-vs-16-bit-audio-test-part-iii.html
Compressed

However, the last 20 year's pop music trend of 'mastering for maximum compressed levels' and to the 'least common denominator of the listening environment' has already assured the audience is no longer listening to what the artist hears in the studio.

A simple means for checking the dynamic range of MP3s can be found here: (http://dr.loudness-war.info/) and the software for checking your own MP3 files via 'TT-DynamicRange 1.4' as downloaded for free from the same site (Upper Right corner links). Viewing the reported data on the site, or just checking your own music files, provides some very interesting trends.

Listening trends in pop music include ...
  • The effect of the 'Loudness Wars' is most evident on more recent 'pop' music from less-well-known artists
  • Even hard rock of earlier decades (eg. The Who) often has much better then average dynamic range
  • The more prominent artists appear to be much more interested in preserving dynamic range to preserve their art
  • If you don't listen to pop of the 80's and later, most of your music is not so badly compressed as you'd thought
It is clear that MP3 itself is only a small part of the problem -- and a technical one at that. If high dynamic range is preserved throughout the process, most of the quality limitations will remain moot. Then, 24 bit lossless files can define the last step (over 16bit/44.1kHz) in achieving a superior technical audio path to the listener -- IF the listener is willing to listen with high quality equipment -- quality headphones or well-designed listening environment vs earbud and a cell phone.

Here are some details ...

Sample Rate (44.1kHz / 96 kHz / 192 kHz)

Digital format recordings (as opposed to analog recordings, eg; vinyl LPs) store samples of the music's analog sound waveform at uniform time intervals. Sampling a single tone sine wave at just 2 points (say the positive and negative peak values) would be the lowest possible sample rate to even partially represent that tone. So, one half the sample rate, known as the Nyquist rate, represents the maximum frequency that can be digitally encoded; even with substantial signal loss.

Bit-Depth (16 vs 24)

Digital format recordings (as opposed to analog recordings, eg; vinyl LPs) store a sampling of the music's analog sound waveform; binary encoded to a pre-defined sound level resolution. 1-bit can save 2 discrete levels, 2-bits can save 4 levels and so on to where 16-bits can encode 65,536 levels. Since 24-bit recording is (24-16=8) bits more, 24-bit recording can represent the analog signal with 256x greater resolution.

Lossy Digital Encoding (MP3)

To save data storage space and streaming bandwidth, digitally recorded music is almost always processed into a 16-bit, 44.1kHz sampled format which is then further processed to remove any "un-necessary" data -- often a 10:1 reduction in storage space -- MP3 files. The result is 'lossy' encoding because the data lost from the original source can never be recovered. Simple listening comparisons between FLAC and MP3 from a CD easily shows the degradation of MP3 encoding.

Lossless Digital Encoding (FLAC)

CDs are encoded in FLAC (Free Lossless Audio Codec) format with the same 16-bit, 44.1kHz sampling but it does not include the MP3 lossy encoding step. Therefore, no data is 'lost' from the original 16-bit, 44.1kHz 'CD-master' stereo tracks.
  • NOTE: Yes, the FLAC format has several optional compression options -- HOWEVER -- all of the compression is LOSSLESS compression; no data is lost or altered from the original analog-to-digital encoding. This means that WAV format files a NOT more accurate than FLAC files. WAV files are simply a more primitive and more direct digital format and so usually larger files. The single difference (other than format) between WAV and FLAC is that the decoding of compressed FLAC might be ever so slightly slower than reading WAV files. This is usually never noticeable.

Studio Formats (24/96 and 24/192)

Recording studios frequently record music source material in 24-bit and either 96 or 192 kHz sample rate. There are three primary reasons for this:
  • The technology can easily deliver this resolution so there is little additional cost
  • Higher bit-depth means higher dynamic range which makes recording and mixing easier
  • Higher dynamic range means expanded range for mixing which reduced the need for compression of tracks
  • With careful use and quality equipment, 96 kHz can deliver nearly 48 kHz frequency bandwidth
These technical values simply mean that recording levels are less critical, the noise floor will be less of an issue, mixing tracks with different recording levels will be easier and mixing large numbers of tracks will be a simpler process.

Mixing Effects

The mixing process creatively assembles all the tracks into the final musical representation the artist intended. The final studio test is control room monitoring in hopefully and ideal acoustic environment -- which is how the artist would like their audience to hear the music. While each individual track may have been EQ'd, leveled, compressed a bit, filtered and processed in other ways, the final mix is the artist's composition. What happens between studio monitoring and a streamed MP3 heard through ear-buds on a smart-phone while riding taking the transit to work can be dramatically different.

Mastering

Originally, mastering was simply the 'live' real-time process of ensuring levels ( above the noise but avoid distortion) when transferring audio to the 'master' vinyl medium for LPs. With the advent of music over AM radio, stations started to actively 'adjust' the levels from LPs to ensure the maximum output levels at all times to counter radio static noise from weak radio signal strength. FM avoided most of this but the trend ultimately led to the 'loudness wars' driven by marketing pressure to promote records by 'being the loudest'. The technical industry responded by building better and more clever compression tools to make that process easier. Starting in the early 1980s, this produced:
  • A Mastering process focused more and more on compressing the music's dynamic range
  • A dramatic increase in actual clipping of music peaks -- adding to the distortion
  • Radio stations adding additional compression and processing -- independent of mastering compression
  • Additional compression artifacts from mass market MP3 encoding schemes
  • A global loss in dynamic range and quality for the audience
A mastering process, focused on bringing out the best in the music, remains a necessary step. It is the heavy compression effect from the loudness wars which distorts both the process and the music.

HRA Wars

The audience, ever hungry for the latest gear, is now (in the mid 2010s) seeking to 'hear' what they believe sounds most like a 'live' performance of their favorite artist. However, a side-effect of modern technology (portable digital players, internet streaming, etc.) is the false expectation that the latest technical numbers increase will always yield a better experience; 24 must always be better than 16 just because the number is higher. Unfortunately, un-biased comparisons are more difficult to achieve than they might seem and subjective claims only obscure reality.

Weak Link

The audio chain extends from performer to playback device and the sound cannot get any better than the weakest link in that chain. If the performance, mixing and mastering achieve what the artist wanted but that source is then filtered, re-leveled, compressed and then played back on a narrow bandwidth device with little dynamic range, the listener will not experience what the artist intended.

"Beats(R)" Audio

The Beats(R) marketing effort is a classic example of attempting to adjust what the listeners hears to resemble more closely one particular artist's style in the studio -- and, equally important, sell more stuff to consumers. No technical specs have been published so this product's primary effect is roughly equivalent to a stereo's Loudness switch - on steroids - which simply boosts the low and high ends of already heavily compressed pop music. Needless to say, this is primarily a mass-marketing effort as opposed to any effort to address underlying audio quality issues. While some listeners may like the effect, it offers no real path toward faithful reproduction of studio recordings.

"PONO(R)" Audio

Neal Young's now-famous HRA pocket-player and music service is different in two respects:
  • There is a substantial investment in new technology and music file delivery service at 24/192
  • The stated goal is to help the audience get closer to hearing musician's studio performance
While the 16-bit vs 24-bit controversy still rages over PONO(R), the second goal of bringing an artists' true performance product directly to the listener is an admirable effort to reverse years of mass-marketing of over-compressed music driven by the loudness wars and record sales.

However, what is not yet available is a clear demonstration of 16/24 bit resolution vs re-mastering of studio product. It has been clearly shown that the number of bits is significantly less important than preserving the dynamic range and quality of the final mix (w/o Mastering).

Frankly, I suspect that PONO(R) is not differentiating between the 16/24 bit change and the difference between a lossless 24/192 studio final mix and a Mastered-for-CD, heavily compressed, lossy encoded MP3. The two different factors (bit depth and mastering) are intertwined and the perceived change for listeners is then attributed solely to the 16-to-24 bit change. The intent is good but such pseudo-science claims would be sloppy at best. Why hasn't a dramatic example of PONO's(R) music capability been released, for free, for comparison by anyone with the technical capability to handle the files?

24/96 FLAC vs 16/44.1 MP3

Given that the difference between 16-bit and 24-bit depth is essentially inaudible, a different approach to observe the impact of the MP3 format is to compare a studio 24/96 FLAC vs a 16/44.1 MP3 of the exact same music where they have the same source. For example, HDTracks(R) sells 24/96 studio source lossless FLAC files. They list the Privateering(R) album from Mark Knopfler and that album was created in his own British Grove studios. Both 24/96 FLAC (from HDTracks) and a 16/44.1 FLAC (from the CD) files can be imported into the Studio One DAW where they can be listened to and analyzed in their native format. (The 24/96 FLAC can be imported directly and then remain in that bit depth and sample rate. The CD's 16/44.1 FLAC can be imported and a 16/44.1 MP3 version can be decoded to 24/96 WAV format which is effectively a lossless up-sampling.)

Psycho-acoustics

While dynamic range, frequency response, distortion, average level, compression and spectral content of audio files can be measured, the much-less-well-defined-or-understood field of hearing perception must rely on poorly defined, over-used terms to describe what listeners hear. Terms like openness, clarity, dark, airy, broad, sharpness, localization, dynamics, reverberation, etc. attempt to translate the technical aspects of the recording's acoustical coloration; and, as with the term 'coloration', they are clearly self-limiting.

Hypothesis: I suspect that a dynamic combination of existing measurements may eventually provide a tractable 'image' of what psycho-acoustics terminology is attempting to describe. This would provide, at a minimum, a meaningful reference point for establishing improved tools for the home-recording artist to improve their recording skills. It might also help establish a more quantitative reference for assessing the improvements of 24-bit studio masters vs 16-bit copies.