Today MQA Audio

But remember your DAC will see it as a 24 bit file with a fair amount of crud - just did a listening test here with my iMAC - hardly hifi - and vocals are certainly less cleaner when you set to MQA passthrough - and Audio Midi Setup confirmed sample word size and bit rate.

Simon-in-Suffolk posted:

But remember your DAC will see it as a 24 bit file with a fair amount of crud - just did a listening test here with my iMAC - hardly hifi - and vocals are certainly less cleaner when you set to MQA passthrough - and Audio Midi Setup confirmed sample word size and bit rate.

I was thinking about this - well - the question of the extra information being stored in less significant bits creating noise.

So my train of thought is this...

16 bits gives a resolution of roughly +/- 33k.  The least significant bit is subject to rounding errors, which gives you your digital SNR of 96db.  So your dodgy digits (+/- 1) are running at about 0.00305%

With 24 bits you get resolution of +/- 18.8m,  but if you then use the 8 least significant bits, as MQA does, to fold in Hidef detail and everything else it does, your dodgy digits give you an error of +/- 256 - which is also .... exactly 0.00305%

So I think this means that for unprocessed MQA files, where they are treated as 16 bit files, with the least significant digits dropped or ignored, then they will be exactly the same (except that the benefit from the MQA anti-smearing treatment to the masters - i.e. they should be a tiny bit better), and have no extra noise.

2. Where they are treated as 24 bit files, with the least significant digits being processed by a DAC, then I am unsure what happens.  It has been said the MQA uses 16 digits unaltered, and the other 8 digits to hold the hidef and interpretation information.  But I read on another forum someone arguing that MQA doesn't necessarily use all 8 bits - it uses as many bits as it needs to until it reaches the noise floor of the master - which might be 13 (80db) or 20 (120db).  And whatever is left is used for the additional data it has 'folded in'.   If that is correct, then the extra data is always under the noise floor, and I'd like to suppose that there would also be some dithering and or filters in the low end digits to minimise any artifacts being created.

Somebody needs to run a course for confused audiophiles.

But I don't think there is anything really as 'anti smearing' .. you can preserve temporal information and bandwidth, which reduces phase distortion or smearing..but you need the hidef processing to recover using a filter...i.e. Your HF noise on the least significant bits is giving you the information to reconstruct your high bandwidth signal... smearing occurs when the signal is bandwidth constrained and or has frequency/phase distortion. Digital audio nearly always uses dithering on the least significant bit to remove quantisation error and makes the reconstructed signal sound more natural..

MartinCA posted:

From Googling around other sites, there is much confusion and contradiction about this whole Tidal announcement.  I put in a support request to Tidal asking them for clarification.  If they tell me something helpful I will relay that on to you.  

This was to do with bit resolution files the Tidal app was delivering after unwrapping.

So,  I got a reply from Tidal.

They said ....  the Tidal PC application does 1x unfolding, and MQA DACs do 4x unfolding.

Erm, right.

I kind of thought that 'folding'and 'unfolding' was a marketing metaphor for what happens, and that the 'unfolding' process wasn't multiple unfolds like from 48khz to 96 to 192 to 384, but rather a single process from 48 to whatever the target frequency is. 

So I don't know what they mean by that, and I suspect the guy who sent the reply didn't either.

However, I put in another support request to Tidal asking them for clarification about what they mean by 1x and 4x unfolding.  If they tell me something helpful I will relay that on to you.  

I think I'll go and listen to some calming music - let me dig out the old Riga Planer.

Folding is marketing speak as far as I can gather  for the effect of decimation and then oversampling

Simon-in-Suffolk posted:

But I don't think there is anything really as 'anti smearing' .. you can preserve temporal information and bandwidth, which reduces phase distortion or smearing..but you need the hidef processing to recover using a filter...i.e. Your HF noise on the least significant bits is giving you the information to reconstruct your high bandwidth signal... smearing occurs when the signal is bandwidth constrained and or has frequency/phase distortion. Digital audio nearly always uses dithering on the least significant bit to remove quantisation error and makes the reconstructed signal sound more natural..

In this context I understood time smearing to occur in some (older?) ADCs where transients are 'smeared' into adjacent time slices, a problem sometimes compounded by brickwall filters, with ringing artifacts created, and sometimes further compounded by remastering. 

But when I mentioned anti-smearing, it was just a passing reference to the preliminary rework of the master files used by MQA to reduce any time smearing and ringing issues.  MQA say that their new master files are better (because of this), and if so, then the benefit will be felt whether or not the downloaded file is unfolded - which was my point really. 

Dithering - yes indeed. It randomises the bit setting.  You could apply it to bits at a higher level of significance though, at the point below which you are in the noise level, if you are are using lower order bits for another purpose.  I was thinking that this would help prevent the folded-in information creating artifacts.

Martin

Martin, need to see more details really... yes smearing can be reduced by enhancing the reconstruction filter in the DAC, and increasing the kernel size of the filter (taps) as well as the design of any apodizing function around the filter. This is what Rob Watts focuses on with his Chord DACs for example.. But this is for playback for any master.

For an ADC encoding the same holds true, but the effect will depend on the quality of the original masters... most MQA albums currently are back catalogue... so at best all I can see is the temporal precision of the master is largely maintained if using a MQA filter on reconstruction as well as encoding... otherwise I can't see what is happening other than comparing methods of downsampling the master for creating a distribution master of the media... and I can't see that MQA adds much if anything to current methods if the encoding and decoding filters aren't matched

MartinCA posted:

MartinCA posted:

From Googling around other sites, there is much confusion and contradiction about this whole Tidal announcement.  I put in a support request to Tidal asking them for clarification.  If they tell me something helpful I will relay that on to you.  

This was to do with bit resolution files the Tidal app was delivering after unwrapping.

So,  I got a reply from Tidal.

They said ....  the Tidal PC application does 1x unfolding, and MQA DACs do 4x unfolding.

Erm, right.

I kind of thought that 'folding'and 'unfolding' was a marketing metaphor for what happens, and that the 'unfolding' process wasn't multiple unfolds like from 48khz to 96 to 192 to 384, but rather a single process from 48 to whatever the target frequency is. 

So I don't know what they mean by that, and I suspect the guy who sent the reply didn't either.

However, I put in another support request to Tidal asking them for clarification about what they mean by 1x and 4x unfolding.  If they tell me something helpful I will relay that on to you.  

I think I'll go and listen to some calming music - let me dig out the old Riga Planer.

'1x, 4x' they could simply be talking about the speed of unfolding?

Unfolding is just a fancy term MQA use for oversampling the decimated digital signal.

The base band of MQA is 48KHz

I am guessing that 1x unfolding or 4x unfolding is the the degree of oversampling or 'unfoldeding'. 

Therefore I am guessing  1x oversampling is 98kHz  or 88.2kHz and 4x is 192kHz (although really that should be 3x...)

I could be wrong - but in looking at parts of how the MQA patent work that potentially makes sense to me.

It also means the software decoders are limited to 96/88.2 which is what we see..

Simon

Dont expect clarity from Meridian about what subversion of MQA playback is in use. They have apparently decided to lump everything related to MQA into an "MQA" marketing bucket, taking no account of which version, which capabilities are in use etc

Simon-in-Suffolk posted:

Alan - you are right - it is not as rich - its ever so slightly too high end in balance ...I was perusing some documentation today on MQA and some analysis of the patents - and apparently it is understood that the higher frequencies that are reconstructed have to be boosted as they become attenuated somewhat when 'folded' with the encoding decimation .. therefore this 'boost' is a subjective enhancement.

Thanks Simon.

This is interesting, but this is no longer lossless audio anymore

I find Rob Watts overengineered attempt to reconstruct the analog waveform more musical, and more transparent.

It's nice to have MQA, but also nicer to have a Chord Hugo to use with non MQA files

I had another note back from Tidal.  It doesn't reveal any more, but the link it provides is informative....

They say...

The software does not decode (unfold) it completely. You need an MQA DAC for that.

In order to listen to the MQA format to its fullest extent, your equipment needs to be MQA capable. Otherwise, you will not be getting the full MQA experience but you will still have better quality than HiFi.

Please read this article as it is informative. http://audiophilereview.com/cd...-on-tidal-rules.html.

The article states that to output of the Tidal App to non-MQA DACs is either 88/24 or 96/24khz.  So I guess that is what they mean by 1x unfolding.    The article also says that MQA DACS can 'unfold' or decapsulate detail to a 384/32 level (with ADC/DAC matching, though it doesn't say that).   I guess that would be 4x unfolding.   But it isn't clear what 2x or 3x unfolding might mean - we can guess, but to be honest, it's fairly meaningless as far as I am concerned.   Of course what the article says may not be correct, but in the absence of evidence to the contrary, I think I'll go with it for now.

The interesting unknown to me at this point in time is what has been done to the MQA master before it is encapsulated i.e. what (if any) improvements are made to the master that pass through even without decapsulation of the hidef detail.  I've come across a number of posts and articles in other places that say that the master is 'cleaned' or 'time smearing is removed' - but I am not sure what actually has been done or whether those notes are confusing what happens before the en/decapsulation process with what happens during that process.   I'll dig around a little and see what I find.

My Meridian Explorer 2 has arrived. And I'm not afraid to use it...

There's something else that is starting to worry me a bit. 

If we are using the PC to send a PCM stream to the DAC via SPDIF, then the resolution of that bit stream is set in the advanced properties setting for the sound card.  So, for example right, up to now, I set my sound card to output 16/88,  88 being a multiple of 44;  which I saw as a compromise between most music from my PC and Tidal which was 16/44, and the occasional higher def file.  But generally I played hidef through my unitiserve.

So now, Tidal is delivering 24/96 resolution to the PC.  Initially I changed the soundcard settings to match that (actually I used 24/192).  And that sounds good (better) to me.  But now I guess all the rest of my 16/44 music is being converted to 24/192 by the PC sound driver as it streams - introducing at least some interpolation issues and I don't know what else .

Has anyone thought this through and how best to manage it (other than investing in a nice new TIDAL and MQA streamer - in fact without buying any new kit)?

There are a couple of new articles that have input from Bob Stuart. They are an attempt to further explain how software decoding works. They also touch on the differences between software decoding, MQA renderers (e.g. AQ DragonFly) and full decoders (DACs).

An MQA encoded file can be played back in four ways; with no decoding, software decoding, hardware decoding, and a combined software/hardware decode.

MQA Decoding Explained | AudioStream
http://www.audiostream.com/con...d#b4g3dmu3V6QltWS1.9

MQA & Tidal - where are we now? | DAR__KO
http://www.digitalaudioreview....al-where-are-we-now/

As is often the case, further useful information can be found in the comments.

Thanks Greg - those two articles (and, as you say, the comments) are really helpful, and help me get my head around all this!

The relatively recent decision to allow software decoding has definitely clouded what was previously a relatively simple message regarding playback. It's not surprising there is confusion.

Now we have seen the Audiostream and DAR articles, I'm hoping it won't be too long before we get one from Computer Audiophile.

and the other interesting thing I have discovered in looking at how MQA works in terms of decimation and oversampling - is that the reconstructed hidef audio (after so called 'unfolding' or what we more usually what we call oversampling)  appears to be  full of aliased frequencies from the original decimation and it assumes those that listen to the audio are not sensitive to aliasing errors... i have no idea whether I am or not sensitive to aliased frequencies in my digital audio - but true hidef doesn't have these aliased frequencies in the content if filtered correctly.

MartinCA posted:

There's something else that is starting to worry me a bit. 

If we are using the PC to send a PCM stream to the DAC via SPDIF, then the resolution of that bit stream is set in the advanced properties setting for the sound card.  So, for example right, up to now, I set my sound card to output 16/88,  88 being a multiple of 44;  which I saw as a compromise between most music from my PC and Tidal which was 16/44, and the occasional higher def file.  But generally I played hidef through my unitiserve.

So now, Tidal is delivering 24/96 resolution to the PC.  Initially I changed the soundcard settings to match that (actually I used 24/192).  And that sounds good (better) to me.  But now I guess all the rest of my 16/44 music is being converted to 24/192 by the PC sound driver as it streams - introducing at least some interpolation issues and I don't know what else .

Has anyone thought this through and how best to manage it (other than investing in a nice new TIDAL and MQA streamer - in fact without buying any new kit)?

 

Adjusting the settings in the Tidal app to Exclusive mode should automatically adjust the output resolution coming from that output device on your computer. 

Just returned from my 20th annual trip to CES, and with a couple of days to digest ... this is what I have to say on this forum.

The Naim Nova sounds fantastic, IMO better then the stack of Naim separates across the hall. 

The Focal Utopia Headphones are WOAD the finest headphones I've ever heard. (Me being an admitted headphone hater). The new Hugo 2 DAC also sounds special.

IMO the future of audio for consumers is frozen in the grooves of LP's and Streaming. (MQA ... and soon many others) Bit perfect digital has been in the past ... and continues to be ... a complete fantasy perpetuated by the naive. The ownership of digital music files on plastic or computer drives will go the way of the steam locomotive. While we have all spent thousands of dollars on this format, I for one am not spending another dime in this direction. 

Say what you want about HiRes audio streaming audio & the technology behind MQA ... this train has left the station and I for one am on board.

Interesting that the Audiostream article linked to above implies that playback of a Tidal Master album on a Naim streamer should result in 24/48 or 24/44.1 playback.  However, of the many files I've tried to playback directly on my streamer they all report back as "FLAC 44.1" with no indication of bit depth. Also, I have yet to find a "FLAC 48" file though I know that several of these albums have a 24/96 master (48 being half of the sample rate).

Bill Allen posted:

Just returned from my 20th annual trip to CES, and with a couple of days to digest ... this is what I have to say on this forum.

The Naim Nova sounds fantastic, IMO better then the stack of Naim separates across the hall. 

The Focal Utopia Headphones are WOAD the finest headphones I've ever heard. (Me being an admitted headphone hater). The new Hugo 2 DAC also sounds special.

IMO the future of audio for consumers is frozen in the grooves of LP's and Streaming. (MQA ... and soon many others) Bit perfect digital has been in the past ... and continues to be ... a complete fantasy perpetuated by the naive. The ownership of digital music files on plastic or computer drives will go the way of the steam locomotive. While we have all spent thousands of dollars on this format, I for one am not spending another dime in this direction. 

Say what you want about HiRes audio streaming audio & the technology behind MQA ... this train has left the station and I for one am on board.

 

Good luck with that! Hopefully, things will improve, but right now, Tidal as my only music source would be a depressing thought. Massive gaps in the available catalogue, inconsistent sound quality, no guarantee of availability as deals with artists and/or recording companies come and go, etc. Thanks, but no thanks! 

Bill Allen posted:

The Focal Utopia Headphones are WOAD the finest headphones I've ever heard. (Me being an admitted headphone hater). 

Must look them up, I thought.  Flipping Heck - at that price they'd better be!

ChrisSU posted:

Bill Allen posted:

Just returned from my 20th annual trip to CES, and with a couple of days to digest ... this is what I have to say on this forum.

The Naim Nova sounds fantastic, IMO better then the stack of Naim separates across the hall. 

The Focal Utopia Headphones are WOAD the finest headphones I've ever heard. (Me being an admitted headphone hater). The new Hugo 2 DAC also sounds special.

IMO the future of audio for consumers is frozen in the grooves of LP's and Streaming. (MQA ... and soon many others) Bit perfect digital has been in the past ... and continues to be ... a complete fantasy perpetuated by the naive. The ownership of digital music files on plastic or computer drives will go the way of the steam locomotive. While we have all spent thousands of dollars on this format, I for one am not spending another dime in this direction. 

Say what you want about HiRes audio streaming audio & the technology behind MQA ... this train has left the station and I for one am on board.

 

Good luck with that! Hopefully, things will improve, but right now, Tidal as my only music source would be a depressing thought. Massive gaps in the available catalogue, inconsistent sound quality, no guarantee of availability as deals with artists and/or recording companies come and go, etc. Thanks, but no thanks! 

+1.  I'm totally open to a new future, but TODAY, my music listening desires cannot be satisfied with streaming solutions only.  Things clearly need to settle in (out?) a bit more.  The quality needs to be sorted out, the hardware needed to obtain that quality needs to be sorted out, and the spotty content from platform to platform needs to be sorted out.

If when it's sorted, sure, sign me up.  I never bought a ton of dvd's, and in 2017 I feel that I don't need to own dvd's or a video server to see any movie I want to at any time. . . . with a few exceptions of course because not EVERYTHING is on Netflix or the iTunes Store, etc.  If we get to this same place with music, great!  Til then . . . 

MarkMcK79 posted:

Adjusting the settings in the Tidal app to Exclusive mode should automatically adjust the output resolution coming from that output device on your computer. 

Thanks Mark, appreciated.  Yes, indeed - I had switched off exclusive mode because I was fiddling about in Media Monkey.  I should have realised!