Meridian says, "it is MQA's position that no relevant audio information was there anyway, so there is no real sacrifice".
But isn't that what is also said about high-bitrate/constant bitrate mp3 files?
I think Meridian is being a little disingenuous here.
Meditations on Meridian's MQA
One of the hassles of streaming hi-res audio files is the sheer volume of data. Of course, that is the price you pay for hi-res; by definition, it has extended frequency response and dynamic range compared to Red Book CD, and that entails many more bits. One of the most interesting aspects of MQA is its ability to code hi-res audio signals in a manner that reduces the number of bits needed to convey the signal. This makes hi-res much more practical.
To provide streaming efficiency, MQA uses filters to separate high frequencies from the lower baseband, requantizes the high frequencies, and cleverly "buries" that data in least significant bits of the baseband signal. Thus, for example, a 24-bit/192-kHz signal can be conveyed as a 24-bit/48-kHz signal. Of course, you must sacrifice any content that was in the very least significant bits of the original file; it is MQA's position that no relevant audio information was there anyway, so there is no real sacrifice. Also, any information in the bands above the baseband that is deemed as nonaudio is discarded. Of course, after streaming, MQA decoding is needed to put the audio blocks back into their proper places.
Reduced file size is one thing, but how can Meridian claim to improve playback fidelity? There's another trick for that. A/D and D/A converters handle the all-important task of digitizing the analog signal, and analogizing the digital signal. We think of digital audio as being immutably simple because even the most complex audio waveform can be represented by a sequence of 1s or 0s. But the job of converting a slice of an analog waveform into a binary word, and back again, is far from easy. For example, A/D converters (especially older ladder-type ones) introduced subtle nonlinearities, and these can be quantified. MQA conveys metadata about the converter used to the output decoder via the bitstream. Using this information, the decoder can correct those defects during playback. Typical playback methods don't do this, hence MQA can claim to improve playback, provided one of its cataloged A/D converters was used to make the recording.
You might be surprised to learn that while MQA is new, some of its engineering roots have been around for many years. For example, the "buried data" technique used in MQA was first described in a paper by Michael Gerzon and Peter Craven entitled "A High-Rate Buried-Data Channel for Audio CD" published in the Journal of the Audio Engineering Society in 1995. This paper described a wonderfully neat way to code auxiliary data into the least significant bits of a Compact Disc. Now, Meridian has updated this technique and applied it to the wholly new problem of audio streaming.
All in all, MQA is another prime example of how smart engineers (and thousands of lines of code) can bend the laws of digital audio. You'll be hearing a lot about it in coming months. MQA has a number of other tricks up its sleeve and in future blogs we'll delve into its technical underpinnings as well as its pros and cons.
See "Meridian MQA, Part 2" here.
- Log in or register to post comments
... Michael Fremer's teakettle is starting to simmer.