“MQA is a philosophy”..John Stuart

The above referenced article and "Archimago's" work confirm what the impulse response measurements posted in Stereophile say directly with actual data. Noise is added by this "codec" to raise the noise floor and mask the low level ripple (commonly and erroneously referred to as "pre ring") in the impulse signal. For those who work with impulse response signals on a regular basis, they don't need a detailed explanation of how MQA degrades the signal. They can see it immediately in the response graphs. Not only is noise added (dither) to adulterate the signal - thus losing net signal precision from effectively 16 or 24 bits down to 14, but the claimed "time error correction" actually adds time domain distortion to the end result. You can see this in the Stereophile impulse response graph easily from the delayed negative going spike. The impulse response of any linear system or approximate linear system is the fullest expression of signal quality and fidelity that we know of today. It is a complete characterization of the system's time (and thus frequency domain for linear systems) domain behavior. One need look no further than the impulse response graphs. They tell the full story in an instant. MQA is not merely a flawed codec. It's a licensing lock box piece of garbage masquerading as an industry standard for digital media distribution.

If MQA were to actually be an ADVANCEMENT for the digital media distribution industry, it would have provided meaningful compression  of digital files WITHOUT ANY PERCEPTIBLE LOSS OF CONTENT INTEGRITY/FIDELITY. We all know that 16 bit and much more demonstrably - 24 bit digital media precision is largely a waste when it comes to real world signals. The full dynamic range that comes with 24 bit systems translates to 140 db!!! That's enough to make your ears bleed! So clearly, there is potential in the marketplace for a coding scheme that scans digital files, records dynamic peaks in the content, and adjusts bit precision accordingly to fully accommodate the individual file's needs before encryption or compression takes place to facilitate more efficient transfer to the intended target. A new industry wide digital standard could place dynamic range information somewhere in a predetermined location in the media content that signals to adaptive encoding/decoding equipment what algorithm to use to fully accommodate the file without padding it with a hole bunch of 1's and 0's that don't change before it is either transmitted or stored on media. MQA could have done this and dispensed with the entire fraudulent "time correction" BS. They would have provided some factual justification for existence in being able to legitimately say - you now have lossless transmission that is more efficient than the current standard. Unfortunately, they lost all credibility when the actual response data showed a degradation of the signal (no longer lossless) while they were claiming WITHOUT ANY PROOF WHATSOEVER that time domain errors allegedly inherent to PCM were being fixed.

Just one small quibble ... The MQA decoders rely on apodizing filters (no pre-ripple, lots after) which Meridian has championed, but they are available without MQA. You can probably find lots at Stereophile written about them. My DAC let's me choose 3 different filters with MQA turned off, including apodizing.

Having said that, after listening it is not the filter I ever choose to use. I find it far too soft.

Best,

E

In real music signals played back on real linear, time invariant audio systems, there is no "pre or post" anything. You have passband, stopband, and transitions between them which can be anything from very steep to very shallow. The only "ringing" that can occur in such systems happens when they are not properly damped such as for transition bands using high order filters. The whole notion of "pre" and "post" ringing has nothing to do with PCM A/D and D/A conversion and everything to do with impulse responses. And if you know anything about digital sample and hold or zero/hold circuits, you know that the spectral content is tightly controlled and band limited.  There is no high frequency ringing if sampling rates are sufficiently high and the sampled content is sufficiently band limited with low order filtering circuits. If you impose band limiting as Craven allegedly did - rolling off frequencies at the upper end of the sampled content,  it is possible to wind up with an impulse response of mush that attenuates both pre and post ringing spectral content while adding substantial phase delay (energy storage) with attendant post impulse oscillations. It's important to remember that we're dealing with continuous time invariant audio signals. There's no Hanning window. No time zero. Do yourself a favor and buy a cheap used 100mhz scope from Ebay or borrow one from a friend. Splice an RCA cable on your stereo and connect your scope to it while playing back music. See if you can find some impulse spikes that are in the microsecond  range or lower in duration. Every now and then, it can be helpful to take a step back or two to look at the big picture. Context is everything and voltage/current spikes don't exist in a vacuum. There's a great big world of energy storage elements from tiny diaphragms in microphones to 200 gram 14 inch woofers and everything in between. None of this stuff responds  in any meaningful way to stimuli that span mere nanoseconds or microseconds. And neither do our ears. Craven and the Meridian gang have had significant challenges to their professional reputations over the years. I'll leave it at that.