MQA Labs Inspira & Endura

Digital audio can sound wonderful today, and while it’s not quite technically possible to achieve ‘perfection’, this clever pair of plug‑ins gets us that much closer!

We live in an imperfect world, and that applies to audio recording and reproduction technologies just as much as anything else. Of course, modern digital audio formats are immeasurably better across myriad technical parameters than all preceding analogue systems, but some critical listeners remain adamant that “even the best digital recordings fail to measure up to our best analogue experiences”. I’m not entirely convinced of that claim, personally, but agree nonetheless that current digital audio systems are not perfect — and probably never can be. Perhaps a better question is whether any of digital audio’s remaining imperfections are actually audible or practically significant.

The words I quoted above come from MQA Labs’ white paper on the technology employed in their pair of closely related plug‑ins that, as far as I’m aware, are unique in addressing a very specific artefact of digital recording that is never present in analogue recordings. Before looking at the plug‑ins in detail, then, I think it would be helpful if we first consider the issue they are designed to correct...

Pre‑ringing

A noted imperfection of most current digital audio systems involves their time‑domain response and, in particular, the pre‑ringing artefacts typically introduced by the linear‑phase anti‑alias and reconstruction filtering employed in pretty much all A‑D and D‑A converters.

Imagine a brief signal transient or audio impulse passing through a regular analogue filter. What comes out of the filter is the original impulse followed by a decaying resonance or ‘ringing’ that’s caused by the way energy is stored and released in all filtering circuits. (This natural phenomenon can be found in all forms of filtering, whether electronic, mechanical, acoustic or biological.)

In the real world, an effect (the filter resonance or ringing) follows an event (the transient impulse). The effect cannot pre‑empt the event as that would go against the progressive unidirectional nature of time. Professor Hawking wrote a very good book about that! However, things can be different in the realm of digital signal processing, and the design of linear‑phase digital filters inherently results in pre‑ringing before the impulse, along with post‑ringing after. I don’t have space here to explain why that happens, so I’ll ask you to just accept it on trust. I explored this topic in some depth within an article about the MQA music format back in 2016 (https://sosm.ag/mqa-format) and if you’d like to know more about pre‑ringing and how it can be corrected, I’d suggest reading that.

There are some good academic arguments that the ear/brain is sensitive to the timing information of audio signals, and that pre‑ringing artefacts interfere with that timing information in a negative way.

Although clearly an ‘unnatural’ artefact in the purest sense of the term, filter pre‑ringing is, I’d argue, inaudible to average listeners in typical listening situations. Digital audio has been mainstream for over 30 years now, on radio and TV, the streaming music player, or in the cinema or nightclub, and I’m willing to bet you’ve never once heard your friends, siblings, parents or grandparents complaining about audible pre‑ringing artefacts! Yet, there are some good academic arguments that the ear/brain is sensitive to the timing information of audio signals, and that these pre‑ringing artefacts interfere with that timing information in a negative way, degrading some aspects of audio recordings, including stereo imaging precision and the perceived realism of acoustic spaces — things that experienced critical listeners are, perhaps, more likely to notice. It makes sense, then, that if pre‑ringing can be corrected it is probably worth doing so, to remove a potentially deleterious digital‑only artefact and make the digital signal path sound more like that of an analogue system.

Interestingly, the extent of FIR filter (pre‑ and post‑) ringing is related to the sample rate, with higher sample rates having substantially shorter pre‑ringing artefacts. This may be why some critical listeners prefer higher‑sample‑rate recordings. Similarly, some prefer the sound of minimum‑phase filtering — which has no pre‑ringing artefacts but has a non‑linear phase response — that many current converters offer as a user‑selected option. Unfortunately, very high sample rates generate huge amounts of data to be stored and processed. Along...