MQA decoding issue

[PeterSt]

2.10c has been uploaded to the 2.10 topic.

This is only needed for the Audio PC. Unless you are used to play (MQA) music on the Music Server PC; then it is for that one (too).

Peter

[PeterSt]

I did download a MQA encoded track "MAGNIFICAT 4. Et misericordia" from 2L but XXHE played it as 44.1 FLAC with either mqa decoding on or off. And the iFi MQA enabled DAC processes it as 44.1

Dear Zheng,

Can you please re-apply all of your testing with 2.10c ?
Quite a lot of things were not correct with the "FLAC header determination" (as seen from Arvind's installation).

Regards,
Peter

[arvind]

Hi Peter,

All working well with 2.10c upto MQA 96, however with MQA 192, I am still getting input sampling rate 96. This is even with new prepared streams. Playback of the MQA 192 streams is ok, but input shows 96.

Best regards,

Arvind

[hudesigns]

Peter,
Thanks! I'll do ASAP.
ZH

[hudesigns]

2.10c has been uploaded to the 2.10 topic.
This is only needed for the Audio PC. Unless you are used to play (MQA) music on the Music Server PC; then it is for that one (too).

Hi Peter,
I just replaced XXHighEnd.exe with the new version 2.10c in audio PC. It seems solid without any issue so far. Next I am going to replace it in music PC too. But I need to ask a question. In the above quote you mentioned audio PC in particular but at the download area where you did not mention audio PC at all. Just wondering.

Also on the MQA hardware decoding front, after consultation with an iFi tech I was able to let my MQA enabled (recent firmware 5.3 update) iFi iDac2 to recognize MQA stream but only upto 96kHz!

Second while playing in XXHE with MQA decoding OFF, I was not able to let the same DAC to recognize its MQA stream so it remains at 48kHz mostly with a few at 44.1

Now you just mentioned that with 2.10c, you resolved some FLAC headings issue, which makes me think that if I re-download some Tidal MQA albums I might see something different. So I am going to update right now in my Music PC and will report back tomorrow.

Zheng

[PeterSt]

2.10c has been uploaded to the 2.10 topic.
This is only needed for the Audio PC. Unless you are used to play (MQA) music on the Music Server PC; then it is for that one (too).

Hi Peter,
I just replaced XXHighEnd.exe with the new version 2.10c in audio PC. It seems solid without any issue so far. Next I am going to replace it in music PC too. But I need to ask a question. In the above quote you mentioned audio PC in particular but at the download area where you did not mention audio PC at all. Just wondering.

Hi Zheng, yes, I have been struggling with that text a bit. I now changed it again.
It went wrong when a 2.10a which was in there at first, was changed into 2.10c.
Thank you for being corrective !

Peter

[PeterSt]

All working well with 2.10c upto MQA 96, however with MQA 192, I am still getting input sampling rate 96. This is even with new prepared streams. Playback of the MQA 192 streams is ok, but input shows 96.

Hi Arvind,

The input to XXHighEnd - as a software player - is correct. This is always the Native Sampling Rate as how it is denoted in the file itself. You can see it like this : the file as how it comes from Tidal first needs to be decoded, which means "expand" it to the first "unfold", if present (native 44.1 and 48 files also exist and they don't have a 1st "unfold" stage). When that 1st "unfold" has been done, it is given to the player (XXHighEnd) which thus sees 88.2 or 96 (and that XXHighEnd is applying that decode step herself may be confusing but say that it happens right in front of XXHighEnd).

Software can never "unfold" more than to 88.2/96, which is why you never will see a higher input to XXHighEnd.

What I noticed in your system is that you set the upsampling to 1x. Thus now e.g. 96 stays that (I think you set it to 2x but with the fx button active but for 88.2 and 96 it comes down to the same - you won't upsample to 192 and beyond.
I would set 16x there, as usual (this is up to you of course, and I am only telling to ensure that you understand the process).

If we regard the 2nd and 3rd "unfold" to be upsampling steps anyway for MQA (and I do) then you now do this yourself. One thing : I did not build a formal "till there and no further" mechanism, or IOW, there is no such thing as allowing you to upsample to the "native" Sampling Rate as MQA depicts it (in the file). You can only set 2x, 4x, 8x or 16x. Well, you just set there what you were used to (and I'd say this is 16x).

Try to keep in mind : while 2nd "unfold" and beyond is an upsampling step anyway in an MQA DAC, you now apply that upsampling step yourself with the difference that you use your own filters (within limits of course). Would you not have a NOS1 DAC but an iFi etc., then in such a DAC filtering is applied which you can also overrule by XXHighEnd to some extent. This is not related to MQA at all and always has been so.
Today though, people may get easily confused by their MQA capable DAC (like Zheng's iFi now), that seemingly working very differently. But all such a DAC does is enable a 1st "unfold" which can be done in software just the same (100% the same) and next apply filters which also can be done in software (but with different filters like always) for further upsampling. It is this latter which you have virtually switched off, by setting 2x only.

For Zheng : IIRC he does not upsample/filter in the first place. So he has set 1x (if all is right). This is everyone's good right and in fact you tell your DAC to take care of the filtering all the way (and would one have a NOS DAC then NOS is used as ever back intended ("filterless")). However, when Zheng sustains this 1x setting (already required to decode MQA by his MQA capable DAC) he is comparing apples with orangeswhen he would compare MQA done by XXHighEnd v.w. MQA done by his MQA capable DAC. Why ? because with 1x in XXHighEnd always, he will compare the iFi filters outside of MQA (no MQA file playing) with the filters within MQA (MQA file playing). He now actually compares two different DACs. Would Zheng upsample 4x because his native iFi can do that for non-MQA files (plays 192 (or can it do 384 ?)) and set to 1x to MQA files, he will compare XXHighEnd filtering with MQA filtering. BUT, would he play MQA through XXHighEnd and upsample 4x he would compare MQA playback through XXHighEnd with MQA playback through the iFi. This latter sounds the best apples and apples to me (I know, it still isn't).

Btw, look at the title of this topic. Funney eh ?
So it is all stupid stuff which is almost impossible to uderstand. A decoding problem ...

Peter

[arvind]

Hi Peter,

Understood it. Even before I read your post, I had set the fx to x16. So all seems ok for now.

Thank you very much for your support.

Best regards,

Arvind

[hudesigns]

Peter,
I did some testings and the findings might be of interest to you. The iFi DAC2 with MQA firmware update detects MQA streams from XXHE ONLY in the following conditions:

XXHE settings:
Volume +0dB
fx=OFF (ON if you leave it at 1x)
upsample=1x (input-output, 96-96)
MQA decoding in settings=ON
MQA logo showing on bottom of window while playing

Only the Tidal MQA albums with logo sticker marked as 192, 176, 96, 88.

ONLY all above conditions are met, iFi iDAC2 will detect it as a MQA stream (with specific indication light). But it DOES NOT perform further unfolding.

----------------

If you adjust or change any parameters above, iFi iDAC2 will detect it as a PCM stream instead (again with indication light) with appropriate sample rate indicated by XXHE output value.

If you do any one of the following examples, you will lose MQA stream:

1. if you turn off XXHE MQA decoding in settings, iFi DAC will see it as either 48 or 44 PCM stream.

2. Play any MQA albums marked as 44, or 48, even with MQA logo on bottom showing while playing, iFi DAC see it as PCM stream.

3. if you touch the volume in any way, iFi will see it as PCM stream.

4. if you upsample in any way, 2x, 4x and so on, iFi will see it as PCM stream with a sample rate as indicated by XXHE output.

5. Also both 2.10 and 2.10c behave the same in this test.

Best,
Zheng

[hudesigns]

Peter,
One more observation is that even with iFi DAC detecting the stream from XXHE as MQA stream, I don't see anywhere that indicate that the DAC is performing second or third stage "unfolding" with MQA albums marked as 192kHz. But iFi clearly states that it is capable of decoding MQA stream upto 384kHz.

Also none of the MQA tracks I downloaded from 2L are played in XXHE as MQA at all, either with MQA decoding on or off in XXHE settings. The little logo always show FLAC while playing and input-output is 44-44. iFi DAC always see it as PCM stream.

ZH

[Rmalits]

Hi Peter,

with 2.10 MQA is working well now in my setup, what's great.
Decoding is done well up to 96kHz.
Thank's a lot for your support and the software improvements!

Kind regards
Richard

[PeterSt]

One more observation is that even with iFi DAC detecting the stream from XXHE as MQA stream, I don't see anywhere that indicate that the DAC is performing second or third stage "unfolding" with MQA albums marked as 192kHz.

Hello Zheng,

I just quoted a kind of random sentence from your last posts because I am not sure how to approach this all. One thing is in my mind though : your iFi isn't behaving as should. But how to prove that with my own software which (apparently) can also do a couple of things not right. But point is : you now have a way inconsistent situation. So to me that tells something (to you nothing, I  am afraid).
It is almost like your iFi can't do the 1st "unfold" but can pick it up like MQA because XXHighEnd now is taking care of that part. All seems "opposite" ...

Also none of the MQA tracks I downloaded from 2L are played in XXHE as MQA at all, either with MQA decoding on or off in XXHE settings. The little logo always show FLAC while playing and input-output is 44-44. iFi DAC always see it as PCM stream.

This one is different; So as we know, I now take the FLAC header as the base and if this header doesn't tell it is MQA, then it won't be decoded as well (by XXHighEnd, I mean). If the header tells in the right way it is MQA, then it stil depends on the content, so XXHighEnd legally (read : "authenticated" (ahum)) decodes MQA ot not.
But 2L already has proven in the past that they don't comply well to header data (XXHighEnd contains exceptions for 2L (but also for many more)), so what if this is wrong now. So at this moment (and before I report back about this) I would NOT use 2L for testing whether MQA works or not.

Also be careful : MQA can be downloaded from 2L. This will NOT work in XXHighEnd for MQA decoding. Only the Preparations from Tidal will ...

Regards,
Peter

[hudesigns]

Thanks Peter!
Actually your comments confirmed my doubt about this new implementation of MQA decode by iFi - that is that iFi's MQA firmware only detects a software DECODED stream as MQA stream WITHOUT further participation of 2nd or 3rd unfolding by iFi hardware. Call it "vapor" ware? Or simply a gross negligence?

Just as a matter of record, I am posting the answer that I received from iFi tech in response to my same question. It makes me even more confused:

MQA - 192KHz files are not playing?
Question: On TIDAL I play an MQA 192KHz file but it displays another sampling rate?

Answer: Typicall those albums are encoded as 48kHz MQA file. Meaning Tidal will perform the decoding to a 96kHz audio stream with the MQA renderer handling the further decoding which this has nothing to do with our products, this is how MQA & Tidal works.

To test that the actual audio stream in TIDAL is 48KHz please select "MQA passthrough", it is 96k if TIDAL decodes MQA, it is 48k on MQA pass-through.

Question: But I thought the file was 192KHz remastered?

Answer: Yes, meaning it was MQA encoded from a 192KHz remaster.

Our MQA implementation handles up to 384KHz MQA sources correctly according to MQA specifications.

Ridiculous as is, his suggestion about use of Tidal's "passthrough" is consistent with what I found with XXHE -- with XXHE MQA decode ON I got 96; decide OFF, 48.

Best,
ZH

[hudesigns]

Hi Peter,
This is just FYI. I just posted more questions to iFi technician.

Hi Alix,

Thanks for your patience. To be honest, this is getting me even more confused! Here are more questions:

1. If I let Tidal do the software decode (as seen in video instruction), where does iFi MQA hardware decode come in?

2. I have both a MQA enabled iFi iDAC2 and a non-MQA Chord Mojo DAC. They both behave exactly the same following your video instruction of Tidal MQA setup - that is with "passthrough" OFF you get 96; with "passthrough" ON you get 48. Where is the advantage of using a MQA enabled DAC like iFi with firmware 5.3 update? Does a MQA DAC suppose to do more hardware unfolding to make some MQA albums to 192 or beyond?

So far all we see is Tidal's built-in software decoding without any participation of iFi DAC hardware decoding.
Please give me some guidance. I am very confused!

Best,
Zheng

Here is the video instruction that iFi is sending to everybody:
 https://www.youtube.com/watch?v=_-rMsoORSbw

[PeterSt]

Here is the video instruction that iFi is sending to everybody:

... At least that looks opposite (upside down) to me.

a. MQA Passthrough, what should it mean according to iFi ? My idea about it : don't decode, pass on to Decoder further down the line (but the video shows opposite).

b. Green light instead of blue (or purple-ish), what should it mean ? -> File is MQA Authenticated, but not signed off by the artist (et al) (but iFi hence video shows = PCM).

About your own text in your previous post I must think a little ...

Peter