- Agree if I can choose not to stream MQA also 2) agree strongly 3) agree 4) this would be the best
My comment was more I “polite” way of saying that I don’t want to express a strong opinion before I have listened for myself to MQA. The observation I made is that if the listening tests are correct the Meridian Apodizing filer does not remove the “Veil” as MQA is said to sound better. Hence as MQA is a kind of filter there is potential for a filter to remove the “veil”.
This is also my observation as I have listened to HQ Player and Chord WTA and been very impressed by the improvement. For example, in a direct comparision the Chord Mojo sounded much clearer than a much more expensive Meridian Burson combo. And I think the description “brittle veil” is quite good to describe the Meridian sound compared to the Mojo. Observe that the Meridian/Burson combo sounded great it just had a “veil” compared to the Mojo
From a scentific point of view it is a little hard to understand why such comparably small errors that the standard filters cause can sound worse than the very large increase in errors caused by MQA. This statement assuming that measurements from Miska et al are representative of the 0-24 kHz range after MQA decoding.
Further I think the pre-ring talk is complete rubbish, what I think can be happing is the missmatch between interpolation filter and sample kernel. As we know a Nyquist sampling causes 0 time error as long as the sample rate is sufficently high (i.e. 120 kHz according to the Meridian paper). Hence given that you sample at a high rate you can use basically a point sampling technique. Hence 0 ringen of any kind, of course such a sampling technique would cause a higher noise level as you would fold noise from high frequencies. But as we know high levels of noise is typically quite nice sounding (maybe why MQA sound good despite only having 13 bit resolution).
Hence there is probably no problems from ADC to solve as Miska claim he has verifed for a set of ADC converters. What may be a problem is the recreation of a time continous analog signal. Again as is well know if we use a Sinc function we can perfectly recreate the original analog signal (potentially with some extra noise due to folding from an imperfect “point” sample kernel as described above). But in all practical solutions a window function is used to create an interpolation filter with compact support. This has been discussed much by Chord e.g. how many “taps” does the interpolation filter need, e.g. how aggresive window function is used. Now this window function does cause something that could be considered time domain “blur”. But probably much less blur than MQA.
Hence maybe MQA gives something nice sounding (maybe 13 bits is enough as LP only has at most “12 bits”) but it does it by severly destroying data in the files (which could be ok for streaming). But if you buy files MQA basically not only force everyone to buy new files, but also if some even “nicer” technique (MQA 2.0) is discovered we may need to buy all music again. My current standpoint is that HQ Player / WTA or other solutions are much more attractive from a consumer point of view, as these techniques improve all our old Music also. And even better is if Roon does software decoding of MQA Tidal streaming and everyone can pick what they like.