MQA arrives on Tidal

[Werner]

For 1x rate originals one could, in the case of old and less optimal ADCs equalise their treble response, or linearise phase response. And if aliasing existed it could be cut away, by getting rid of the top treble altogether. But these are mastering steps, not requiring a double ended system at all.

If one really wanted to do something at the decode end one could instruct the DAC to use a different reconstruction filter depending on the ADC type. MP apodising for a half band LP one, and vice versa. This would tie in with Meridian lore.

None of this fixes the mythical time beast, and MQA triangles and origami are not used.

 

[onlyconnect]

I'd be surprised if people here streaming MQA didn't know about the effect of the MQA pass through box, it's been mentioned quite a few times.

Does everyone else get the same behaviour as me, that I get the blue light on the Explorer 2 MQA DAC irrespective of whether passthrough is enabled?

It's intriguing as people have said that changing this setting makes a big audible difference. But the blue light means it should sound the same, at least if you have an MQA DAC.

Tim

 

[Mr_Sukebe]

Now on my free Tidal trial. Would have been online earlier, if my router hadn't had a hissy fit.

Have to say that I'm a little surprised at the level of noise here, e.g.
- Should be free. Just why? Do you people beat the doors down on Dolby labs for all of their software formats, or Sony for SACD and part of CD? Clearly MQA is a competing product. If it's beneficial to the industry, they'll vote with their money, end of story
- "Secret sauce". Again, so what. Did you guys moan when MS released Windows 10 that they hadn't fully disclosed what the code did? For that matter, do you ask for full breakdowns by Mercedes on their fuel air mixture in their injector systems? Just accept that it's Intellectual Property, that they're as entitled to it as other companies and be thankful that MQA have been more forthcoming than most in trying to explain at least at a high level what it does and why.
Have the naysayers even tried it for any extended period of time?

Now, I'm going to spend some time listening to some music I know well. Warming up with the Division Bell, sounds pretty good so far.

 

[onlyconnect]

In the 70s I was happy to pay Dolby licensing fees to greatly reduce Cassette Tape hiss. I have no philosophical issue with rewarding MQA for this innovation in sound , which offers me real musical benefits.

I wasn't; at least, I understand the point you're making, but was shocked by how much recording and playing back without Dolby B improved the sound. Of course with a pre-recorded cassette you were stuck with it!

Tim

 

[onlyconnect]

Do you people beat the doors down on Dolby labs for all of their software formats, or Sony for SACD and part of CD?

Actually I think SACD was in some senses a bad thing for the industry. Leaving aside the controversial question of whether DSD is better or worse than PCM, having an industry war over 5.1 and more capacious stereo formats held us all back.

Tim

 

[SteveS1]

Now on my free Tidal trial. Would have been online earlier, if my router hadn't had a hissy fit.

Have to say that I'm a little surprised at the level of noise here, e.g.
- Should be free. Just why? Do you people beat the doors down on Dolby labs for all of their software formats, or Sony for SACD and part of CD? Clearly MQA is a competing product. If it's beneficial to the industry, they'll vote with their money, end of story
- "Secret sauce". Again, so what. Did you guys moan when MS released Windows 10 that they hadn't fully disclosed what the code did? For that matter, do you ask for full breakdowns by Mercedes on their fuel air mixture in their injector systems? Just accept that it's Intellectual Property, that they're as entitled to it as other companies and be thankful that MQA have been more forthcoming than most in trying to explain at least at a high level what it does and why.
Have the naysayers even tried it for any extended period of time?

Now, I'm going to spend some time listening to some music I know well. Warming up with the Division Bell, sounds pretty good so far.

I don't think you've read much of the thread if you think that is all people are concerned with. The Mercedes analogy doesn't work for this, as the questions around what's happeniing here relate to access to content and the process of encoding. More particularly whether it's an attempt to monopolize access to certain content.

 

[HarryB]

I have Mqa dac here now and the set up is such that you don't have to agonise over whatbbox is ticked etc. It just sorts it. Sounds terrible

Sounds terrible?

 

[Jim Audiomisc]

The thing is that to my ears MQA sounds better than 24/192 so why should they waste bandwidth and get lower SQ?

As has been repeatedly explained already: You don't need to "waste bandwidth" anyway. You can get much the same reductions in stream rates, etc, by non-MQA means. And do it *without* losing or altering the HF or other details of the music. Also do it *without* needing any special DAC or 'decoder.

So turn around your question and ask why you should have to lose details and use special kit whe the only reason for doing so is that you're not being given a choice for commercial reasons?

 

[Jim Audiomisc]

When I say MQA is "effectively" lossless that is my interpretation of MQA's process. The key observation to consider is that real music has only low amplitude ultrasonic content which can therefor be encoded with fewer bits than full range baseband signals without loss of fidelity. So yes, there is perceptual compression going on, but no ultrasonic musical content is lost or encoded at lower resolution than full bit rate PCM. I would describe that as "effectively lossless".

You are making a series of optimistic presumptions about hidden details. The MQA system can be expected to lose some of the HF details *and* alter the output as a result of adding anharmonics at lower frequencies than the input that prompted them. None of that seems a fair use of "lossless" to me.

The difficulty is that we can't currently assess how much it is altering the result and exactly how it differs from the source material. We can't find out if you or I would prefer the source or not to the MQA.

 

[Jim Audiomisc]

OK. I see.

I think MQA does go a long way in sorting out the blurring.

Cheers!

Actually, I doubt that. The folding process tends to generate anharmonics at *lower frequencies* that the HF that caused them to be 'folded down'. So far as I can tell, these aren't then removable during 'unfolding'. That means any edges or spikes may well get added LF. Which will tend to broaden them in time.

This is one of the areas where I think the language being used is being muddled. The results may be 'time aligned' - what engineers might think of in terms of 'uniform group delay'. But you can do that with non-high-sample-rate material anyway without all the MQA handwaving.

 

[Jim Audiomisc]

Another bit of my thinking...

A few years back a pal of mine (who's ears are rather better than mine!) went to an AES dem where an 8-bit - yes that's 256 discrete sampling thresholds! - piano recording was played. It sounded bad, as the notes decayed huge distortion could be heard. Then they played 8-bit with proper dithering, the result was some background noise, but a surprisingly reasonable piano sound decaying into the hiss/noise.

So for playback, 16 bits could well be more than enough.

This reminded me of something which is, perhaps, amusing in this context.

IIRC some years ago Bob Stuart published a white paper to explain what sample rate, etc, he felt was necessary. In effect, carefully noise-shaped 96k/16bit would cover this.

It would also probably flac to stream rates and file sizes that compare OK with MQA. But, of course, be playable without any need for MQA decoding or having had changes made to it a la MQA.

 

[radamel]

As has been repeatedly explained already: You don't need to "waste bandwidth" anyway. You can get much the same reductions in stream rates, etc, by non-MQA means. And do it *without* losing or altering the HF or other details of the music. Also do it *without* needing any special DAC or 'decoder.

So turn around your question and ask why you should have to lose details and use special kit whe the only reason for doing so is that you're not being given a choice for commercial reasons?

You don't have to use special kit as Tidal HiFi allows for MQA decoding up to 24/96. Perhaps it's not the full MQA experience but it's still quite an improvement in terms of SQ IMO vs Tidal redbook streaming.

About losing details what do you mean exactly?

 

[Mr_Sukebe]

I don't think you've read much of the thread if you think that is all people are concerned with. The Mercedes analogy doesn't work for this, as the questions around what's happeniing here relate to access to content and the process of encoding. More particularly whether it's an attempt to monopolize access to certain content.

I believe my comments are bang on, and responding with your assumptions on the circumstances is just lazy. If you have a specific comment to make, please do so.

 

[Jim Audiomisc]

For 1x rate originals one could, in the case of old and less optimal ADCs equalise their treble response, or linearise phase response. And if aliasing existed it could be cut away, by getting rid of the top treble altogether. But these are mastering steps, not requiring a double ended system at all.

If one really wanted to do something at the decode end one could instruct the DAC to use a different reconstruction filter depending on the ADC type. MP apodising for a half band LP one, and vice versa. This would tie in with Meridian lore.

None of this fixes the mythical time beast, and MQA triangles and origami are not used.

Yes. It is a shame that people don't seem to understand digital audio well enough to realise that the result comes from the combination (convolution) of the ADC and DAC functions. *And* that the classic 'sinc' shape was chosen because - once the recording has been made - it 'does nothing'.

i.e. Using a DAC with a sinc filter shape simply gives you the waveform as defined by the recording. Implicit from the start has been that it is the responsibility of those making the recording to define what you should get. The sinc DAC was chosen to allow that.

 

[Stunsworth]

Does everyone else get the same behaviour as me, that I get the blue light on the Explorer 2 MQA DAC irrespective of whether passthrough is enabled?

It's intriguing as people have said that changing this setting makes a big audible difference. But the blue light means it should sound the same, at least if you have an MQA DAC.

I don't have an MQA DAC, so I'm relying on software processing. Software processing only does a partial 'unpack' to 88.2 or 96k. I wonder if with an MQA DAC with the box unticked that initial unpack is done by the app and the second stage is being done by the DAC? Just a guess, probably wrong.

 

[radamel]

Actually, I doubt that. The folding process tends to generate anharmonics at *lower frequencies* that the HF that caused them to be 'folded down'. So far as I can tell, these aren't then removable during 'unfolding'. That means any edges or spikes may well get added LF. Which will tend to broaden them in time.

This is one of the areas where I think the language being used is being muddled. The results may be 'time aligned' - what engineers might think of in terms of 'uniform group delay'. But you can do that with non-high-sample-rate material anyway without all the MQA handwaving.

Have you actually listened to MQA with a good MQA enabled DAC in a good stereo setup?

 

[Jim Audiomisc]

You don't have to use special kit as Tidal HiFi allows for MQA decoding up to 24/96. Perhaps it's not the full MQA experience but it's still quite an improvement in terms of SQ IMO vs Tidal redbook streaming.

About losing details what do you mean exactly?

The software is 'kit' in this context as you can do the same things regardless of where in the chain the decoding is done.

The MQA processes *don't* ensure you lose no details at all. They essentially aim at noting down the details which MQA thinks 'matter' (from their belief in the timing effects and which aspects they think are important). This is a judgement made by the encoding, for better or worse.

The folding also tends to 'add' things. It them becomes an argument about angels dancing on the head of a pin if that is actually losing some details because some parts of the waveform are being altered.

 

[SteveS1]

I believe my comments are bang on, and responding with your assumptions on the circumstances is just lazy. If you have a specific comment to make, please do so.

Bang on the mark? Your comments add nothing. You don't seem to have any more idea of what you are listenng to source wise than anyone else. My comments are already up thread having listened to it and moved on to try and understand what is being presented. It sounds good, that's neither here nor there without knowing what I'm listening to. If you have something to add to that, let's have it.

 

[radamel]

The software is 'kit' in this context as you can do the same things regardless of where in the chain the decoding is done.

The MQA processes *don't* ensure you lose no details at all. They essentially aim at noting down the details which MQA thinks 'matter' (from their belief in the timing effects and which aspects they think are important). This is a judgement made by the encoding, for better or worse.

The folding also tends to 'add' things. It them becomes an argument about angels dancing on the head of a pin if that is actually losing some details because some parts of the waveform are being altered.

If you compare MQA 24/48 or higher to redbook I don't think you can say that you're losing details

And I think one should keep in mind that in streaming it's either MQA or redbook or the ubiquitous MP3.

 

[Jim Audiomisc]

Have you actually listened to MQA with a good MQA enabled DAC in a good stereo setup?

No. Nor are my questions or concerns regarding what the result sounds like. If I were able to get either:

1) Copies of high rez *input* to MQA encoding and the resulting MQA output - no other alterations

and/or

2) Full info on the MQA encode and decode processes

I'd be happy to do tests by listening as well as by analysis and measurement. I suspect the analysis and measurements would tell me more, but that's a guess.

But failing that, I'm not particularly interested in listening to it. I'm quite happy to accept that others prefer it to, say, Audio CD. That's fine. My curiosity about that is *why* they prefer it. Simply listening in itself won't tell me if its an 'effect' or not

In addition, I can't claim to be 'golden eared', so if I heard no difference or decided I preferred a plain high rez source, I doubt that would prove anything to anybody. Better that they make their own mind up on that. The problem being, how many have the real *input* version to compare with?