@dannybgoode
I stay out of discussions like this on business questions, but I’d like to point out that none of the streaming services are profitable either. Like most of them, MQA is just entering its 7th year and all of them are effectively in start-up modes where growth is more important than profitability. You run new businesses by working on invested cash, and then working at a loss to grow the business. Profitability is a later-stage goal.
I mentioned 5x “mqa-cd”. Of couse the 16/352.8 comes from a mqa-cd. And guess what. It’s a 16/44.1 file, it will then unfold to 24/88.2 like any 16/44.1 would. And then it’s upsampled (“rendered”) to whatever the dac is capable of. 16 bit files are normally reserved for 44.1 and 48 as those have no hi-res information, but for mqa-cd they make an exception producing a lossier mqa. I explained all above.
The difference is that on an mqa cd all hi-res mqas are 16bit only so they are lossier than the online version. It’s not that hard to see that the encoder is LOSSIER for mqa-cd then for regular mqa.
MQA and streaming services all have the major record labels as investors keeping them afloat. The labels are quite willing for the streaming services and MQA to have losses, as the losses of the streaming services are in large part due to the royalty payments the streaming services pay to…the record labels.
1 Like
You’re the one who said they were making a success of things 
Anyway, it is not the staggering losses that catch the eye rather the distinct lack of revenue. They are only bringing in a tiny amount of money regardless of the sums they are haemorrhaging.
I fully understand how tech companies are funded and grow. I also understand how many fail.
Qobuz and Tidal at least have a decent top line coming in. MQA does not…
1 Like
No, that was @Chrislayeruk.
I don’t have too much faith in the divinations of people who try to read tea leaves on Companies House though.
I do it for a living 
2 Likes
So here’s the problem. For a 16 bit MQA file we have 15 bits for music and 1 bit for the control stream. Just fitting 0 - 44.1 Hz into the 15 bits already makes the file “lossy”. Now, to add 44.1 - 96 or 192 or 352, etc., more bits have to be dropped re the 0 - 44.1 range to make room for the ultrasonic frequencies. Do you think it sane to drop bits from the audible range, so ultrasonic (inaudible) frequencies can be accommodated?
2 Likes
We have ~16 bits for music, the control stream is buried in pseudo random noise (presumably part of the dither). So it doesn’t eliminate the bit from data usage. Also there will be noise shaping so the dynamic range is greater than 16b. Some part of the ultrasonic region is retained and coded into the control stream (based on Bob’s blog) but I won’t discuss what I think further since I don’t know for sure.
No. This has been tested thoroughly. One bit (I forgot whether it’s the highest or lowest bit) in an MQA-CD file is not music - it is the bit that contains the stream for authentication, pseudo hi-res sample rates, etc. There is only 15 bits for music – it definitely is not buried in pseudo random noise. No music is encoded in the control stream. The rest of what you say, e.g. on dynamic range makes no sense (it’s dithering that increases the perceived dynamic range, not mqa encoding.)
1 Like
Well, I’ve designed similar things. I do this for a living.
You’d do well to read, and understand, the MQA patents, providing you have enough DSP background. Also read the Gerzon-Craven papers on burying data in pseudo-random noise. Craven also is the designer of MLP and of the MQA encoding process, and he’s quite expert at this.
Regarding your new comment, as I said, it’s noise shaping that increases dynamic range.
1 Like
Thanks, but I have seen the patents. IIRC, they discuss only 24 bit MQA; I don’t recall seeing one for 16 bit MQA. FWIW, this doesn’t mean that MQA is implemented exactly like it says in the patent.
A bit of an obscure acronym that needs context to resolve. To save anyone the web search bother MLP = “Meridian Lossless Packing” (convenient for DVD audio) not Google favourite “My Little Pony”…
5 Likes
16b is discussed to the extent of deriving a viable 16b version from the full 24b word, for use in the situation when a DAC doesn’t support MQA. Bob Stuart’s blog has further comments on the MQA-CD process but not a great deal.
That’s certainly true about not being implemented exactly as in the patent.
CDs only support 16/44.1.
Thank goodness we have open source compression schemes.
1 Like
Most of the problems with MQA is that, even if the enhancements to some music sound more pleasing to some degree (though the last MQA I tried yesterday, Jack Johnson, sounded like it had been put through a 3D plug-in on low settings, as the vocals were sucked out and there was a false sense of space); each one of their claims either contradicts their other claims or actual scientific fact, or what they do the total opposite of what they claim, such as the batch-processing of pre-MQA music on TIDAL, or the fake MQA (16/44.1 MQA albums bit-perfect identical to the originals).
People can argue to death about it, post appeals to authority arguments, but (to borrow from an (in)famous celebrity) facts don’t care about your opinions.
I reckon that, if anyone wants MQA, just install one of the 3D plug-ins and use that on low settings for the same effect as what they appear to be doing with recently released albums.
2 Likes
I care not to defend MQA. But look at the above from a mathematical perspective. MQA almost certainly does not steal the LSB 100 percent of the time. Such would create an 88.2 kbps data stream – which should be far greater than 16 bit MQA requires for authentication and upsampling filter configuration.
Even if MQA were to truncate and replace every 16th bit, each LSB has only two possible states – 1 or 0 – so still would have a statistically 50 percent chance of being the same state as the LSB in the original 16 bit music sample. In the end, what effect would that have? Assuming a flat (non shaped) noise floor, reducing signal to noise ratio and dynamic range by 3 dB? Not a big loss.
AJ
Grateful Dead workingmans dead mqa 96 had a similar problem with a very strong reverb… the 50th anniversary mqa 192 did not.
1 Like
Yes, not a big loss. But also, no benefit!
The information in the control stream is repetitive. This allows the MQA files to be truncated, sliced, etc. and still work. Also, where there is corruption, out goes the “blue” light, letting the listener know something is wrong (you can pinpoint it).
2 Likes
Yes, exactly. The musicians rarely have a say in the “authentication” process. Even more rarely than the recording/mixing/mastering engineers. How can that be a good thing?
5 Likes