- #1
- 10,925
- 3,800
Hi All
Some may know I am an audiophile.
A few years ago, with a lot of hype and no small measure of BS, a new format appeared, called MQA. It created a lot of controversy for various reasons like the need for a special decoder, watermarking of the audio, bit stacking, and a special light on the DAC that came on with MQA material supposedly indicating you are getting what the audio engineer intended (
). To be blunt, I found the marketing around it somewhat obnoxious. I did investigate how it worked, but the full detail was hard to find. I liked it for what it's worth, but some didn't. That's nothing new in high-end audio.
Anyway, they are now bust. IMHO, it is a lesson in how marketing BS can destroy what is not a bad idea.
Now it is a bust, a paper has been published with more of the technical detail of MQA, which wasn't easy to find previously:
https://www.aes.org/e-lib/browse.cfm?elib=20456
It is an updated version of a previous one that had less detail.
The reason I posted it is that MQA stated (I think in Stereophile) that well over 99% of the material, when they chop off and dither the noisy bottom bits, there is nothing above 48khz left, so it is just a normal band-limited signal - but band-limited using a sneaky method and can be transmitted at 96k using a common compression algorithm called FLAC. FLAC has an interesting feature. If, say, a 24-bit source is used and you set the lower 8 bits, to zero, then it compresses to the same size as if it was 16 bits. It is called bit freezing. No sneaky origami, etc., is required; use FLAC with bit-freezing, and you have MQA without the BS - just an unusual way of band-limiting material distributed at 96k that supposedly reduced time smear. It doesn't matter what the sample frequency was before - these days, masters are often done in 2xDSD or even 4XDSD it is easy to convert it to 96k with just the higher frequencies that acually contain noise attenuated a bit. For the very few cases where some musical detail is chopped off, you go to 192k Flac. Sure, the material above 24khz is slightly lowered - but that is way above audibility.
If they had done that, you would not need special decoders, watermarking the audio, this authentication stuff lighting up a stupid light and all the rubbish around it, i.e. no BS. It might have succeeded then.
It is an example of how a good idea can be ruined by bad marketing.
Thanks
Bill
Some may know I am an audiophile.
A few years ago, with a lot of hype and no small measure of BS, a new format appeared, called MQA. It created a lot of controversy for various reasons like the need for a special decoder, watermarking of the audio, bit stacking, and a special light on the DAC that came on with MQA material supposedly indicating you are getting what the audio engineer intended (
). To be blunt, I found the marketing around it somewhat obnoxious. I did investigate how it worked, but the full detail was hard to find. I liked it for what it's worth, but some didn't. That's nothing new in high-end audio.Anyway, they are now bust. IMHO, it is a lesson in how marketing BS can destroy what is not a bad idea.
Now it is a bust, a paper has been published with more of the technical detail of MQA, which wasn't easy to find previously:
https://www.aes.org/e-lib/browse.cfm?elib=20456
It is an updated version of a previous one that had less detail.
The reason I posted it is that MQA stated (I think in Stereophile) that well over 99% of the material, when they chop off and dither the noisy bottom bits, there is nothing above 48khz left, so it is just a normal band-limited signal - but band-limited using a sneaky method and can be transmitted at 96k using a common compression algorithm called FLAC. FLAC has an interesting feature. If, say, a 24-bit source is used and you set the lower 8 bits, to zero, then it compresses to the same size as if it was 16 bits. It is called bit freezing. No sneaky origami, etc., is required; use FLAC with bit-freezing, and you have MQA without the BS - just an unusual way of band-limiting material distributed at 96k that supposedly reduced time smear. It doesn't matter what the sample frequency was before - these days, masters are often done in 2xDSD or even 4XDSD it is easy to convert it to 96k with just the higher frequencies that acually contain noise attenuated a bit. For the very few cases where some musical detail is chopped off, you go to 192k Flac. Sure, the material above 24khz is slightly lowered - but that is way above audibility.
If they had done that, you would not need special decoders, watermarking the audio, this authentication stuff lighting up a stupid light and all the rubbish around it, i.e. no BS. It might have succeeded then.
It is an example of how a good idea can be ruined by bad marketing.
Thanks
Bill
