ענו לי מ-MQA. אז להלן השאלות שלי ובהמשך התשובות שקיבלתי:
===========================================
Dear Madam or Sir,
I have a few questions I hope you can kindly help me with:
#1.
The Wikipedia article about MQA and other references to MQA elsewhere (e.g., from the
Wikipedia article about Tidal) explicitly and repeatedly say that the MQA format is lossy. However, my understanding of an
explanation by Bob Stuart and others (e.g.,
Hans Beekhuyzen) is that when using an MQA decoder, the decoding is not lossy, and the full, bit-perfect original file is restored. Yet, in his
"Origami" explanation, Bob Stuart mentions "lossless" only in relation to buying "B" under "A", but not "C".
Question #1 is: Is MQA with a decoder lossy, even slightly, or fully bit-perfect?
#2. The said Wikipedia article about MQA says that when "MQA-encoded content... played back on systems ... without an MQA decoder..., the resulting audio has easily identifiable high-frequency noise occupying 3 LSB bits, thus limiting playback on non-MQA devices effectively to 13 bit". My understanding from the other sources is that the folded content is stored in a 24/48 or 24/44 .wav file and then losslessly compressed using FLAC.
Question #2 is: Are encoded MQA files stored at less than 24/44 .wav before applying the FLAC encoding? How much truth is there in the 13-bit statement?
Question #3 is: Is MQA useful when the original file is below 24 bit and/or at/below 48 kHz?
Question #4 is: When playing an MQA stream in Tidal, is the stream fully decoded to the bit-perfect original file, or just partly unfolded, or played in compatibility mode without decoding?
===========================================
ולהלן התשובות:
===========================================
Dear Amir,
Sorry this is so late, but it has been another CRAZY busy week at MQA!
We don't control the current Wikipedia page on MQA -- on the contrary. Often with new technology there will be individuals who's intent is to disparage or defend a status quo - and that is the case here. We suggest you ignore it as a source of information on MQA.
Instead, here are some authoritative sources that you might find interesting.
A Hierarchical Approach to Archiving and Distribution This AES paper describes the background thinking behind MQA and includes in-depth explanation of the novel digital sampling and why it is not 'a codec'.
The Gentle Art of Dithering This paper covers several topics relating to precision, high-resolution, signal chains and time-domain.
If these are too technical, then the topics you raise are answered in MQA's blog, e.g. try these pages on BobTalks:
http://www.bobtalks.co.uk/blog/insight-series/mqa-a-new-paradigm/
http://www.bobtalks.co.uk/blog/mqa-philosophy/what-is-mqa/
http://www.bobtalks.co.uk/blog/mqaplayback/origami-and-the-last-mile/
http://www.bobtalks.co.uk/blog/insight-series/mqa-problem-solved-2-high-resolution/
Now, to quickly answer your questions:
#1. MQA aims to deliver exactly the sound that was heard in the studio. Since A/D and D/A converters are lossy and different, what matters is consistently and accurately delivering the analogue and so, in a sense, the question is out of framework.
#2. The 13-bit assertion has never had any basis in fact. The MQA file is generally 24b - it can be 16b for original digital recordings made at 48 or 44.1 kHz 16b.
#3. MQA is hierarchical and can be used with digital sources of any original bitdepth and sample rate. To date, 44.1kHz and 16b is the lowest data-rate source we distribute.
#4. The Tidal App has options to engage a Core decoder. That means it can either pass the MQA bitstream undecoded or after unfolding once to give the 'Core'. That signal is one of the studio-approved presentations. Further unfolding can be done using an MQA Renderer.
We hope these give a deeper understanding.
Please stay safe!
Regards,
Ken Forsythe
Director, Market Development, MQA