Bob Stuart, creator of MQA, talks in detail about this revolutionary British technology that sets a new standard in capturing, delivering and reproducing digital audio.
Insight – High Resolution (Intro)
High resolution is an experience, not a specification …
In common audio parlance, ‘High Resolution’ has incorrectly become associated with recordings that use higher-than-CD data rates (i.e. PCM sampled higher than 44.1 kHz or using more than 16 bits).
These simplistic definitions are stuck in the digital domain and tell us nothing about resolution in sound – it only tells us the size of the digital ‘container’ used to distribute information. For human listeners, ‘Resolution’ means to ‘Resolve’ or ‘Separate’ individual details or elements. For sound, this happens in the time domain and it follows that ‘Resolution’ must be defined in analog (since we listen to sound on air) – after that it’s ok to consider how any recording or distribution method compares. (Read more on AES Soundboard)
‘High Resolution’ then becomes an attribute of a complete system in the analog domain (from microphone to loudspeaker). High resolution can’t be guaranteed by sample rate or bit-depth. We must instead consider: Bandwidth, Dynamic range, Temporal blur, and dynamic stability (noise modulation).
The confusion around ‘high-resolution digital audio’ started when people noticed that PCM at faster than CD-rates, could often sound better. This led to an arms-race where increasing sampling-rate to 96, 192 or even 384 kHz could provide sonic benefits – albeit with disproportionate increases in data rate. Similarly, some believe that increasing bit-depth from 16 to 24 or even 32 bits also provides improvements – again with proportional cost in data. There’s no problem in principle using lots of data in an archive, but it is unethical to advocate inefficiency born of lack of understanding.
More data does not guarantee delivering more information.
A file cannot be High Resolution, nor can the container inform us about the sound. To experience resolution, we must play it back in the analogue domain.
MQA goes beyond the flawed measures of digital containers (data rate, losslessness) directly to a ruthless (but ethologically correct) end-to-end analogue definition, modelled on the most transparent sound distribution we know – which is the air around us. Common digital metrics of sample-rate and bit-depth are almost meaningless, instead, we focus on end-to-end delivery with minimal time ‘blur’ and maximum ‘clarity’ of dynamic range (by reducing pervasive errors such as modulation noise).
We have established that the sonic benefits of even the highest sampling rates can be significantly exceeded by using a more modern approach to digital sampling (we call it encapsulation). Combined with an ingenious hierarchical encoding scheme, MQA can deliver all the information at appropriate (lower) rates; for example, using 1.5Mbps to safely render all the sound in a 352.8 kHz 32-bit master (22Mbps). This is not a lossy process, just appropriate efficient packing.
It isn’t necessary to ship a bottle of wine in a 3-gallon carton.
Based on our systematic analysis of more than 5 million 24- and 32-bit songs and detailed measurements of the entire chain (from studio microphone to listener, via A/D and D/A converters, studio workstations and playback systems), MQA has shown that a strong driver of higher bit depth comes from overcoming design defects or misunderstanding. (Reference coming soon).
Our perspective on high-resolution is no longer a matter for opinion or debate.
The MQA system bypasses many points of failure by carefully managing the signal all the way from the studio to your playback. More definition and clarity for less cost.
Since our goal is precision in analogue, we use an array of innovative techniques including hierarchical sampling, fractional-bit coding, end-end subtractive dither and platform-specific D/A rendering management. Again, it is meaningless to compare MQA using inadequate metrics such as sample-rate, bit depth or even intermediate lossless criteria.
high resolution is an experience, not a specification
Along the road, academic researchers have considered 7 different theories in the frequency, time and amplitude domains, about why higher sample-rates or bit-depths might give a better sound with digital audio (and two of these have been comprehensively proven to be incorrect). (More on this in the next post).
MQA has shifted the perspective and frame-of-reference so we need only consider how to deliver the highest sound quality with astonishing clarity and transparency. By designing for human listeners (instead of scientific instruments) and for real music (instead of artificial signals) we were able to put all the available power of digital delivery into a human-appropriate system that values time over frequency, that preserves micro-dynamics by enforcing a lower and stable noise-floor, that identifies the whole music and delivers it intact, but that does not waste energy on artefacts of technology.
Summarizing then, high-resolution is an experience, not a specification – the best 100 Mpx camera fails to deliver with a dirty or out-of-focus lens or on a misty day.
More coming soon …