It can process super-resolution to provide high-quality audio output with a consistent 24 kHz bandwidth and a 48 kHz sampling rate from a variety of sources that produce audio signals with bandwidths ranging from 2 kHz to 16 kHz as input.
This is a little different, I think, but not entirely interesting for most audiophiles. At best upsampling was curve fitting.
This is something you might use to reduce voice over IP (VoIP) bandwidth needs.
For audiophiles much has changed since the CD player. First, in all areas, network bandwidth is just so much better. Your basic 2.4 GHz wifi router has 10x more bandwidth than needed for 96kHz audio, so saving bandwidth is no longer very important. For evidence see the rise and fall of MQA.
Second, DAC’s this century are just so much better than their predecessors at handling low resolution (CD) rate music. The need for 96 kHz and higher signals is just not as compelling anymore.
If I cared about AI, it would be to use AI to go from 44kHz/16 bit to 96kHz or higher. The examples of AI being used to improve video and images is pretty amazing. With the right training, available by taking existing 96kHz recordings, downsampling and letting AI learn the difference, this could be very interesting....
or it could sound just like 44kHz recordings.
Maybe this would be more significant improvement in historical recordings?
I remember when DSD came out and claims of it’s superiority also came with evidence of fraudulent re-eq/re-master work which accounted for much of the claimed improvements.
Overall I think this would be closer to the colorized movies experiments. :D We’ll see.
The good news IMHO here is that this could very much be democratized. We could apply it to our own music libraries or perhaps apply these algorithms via an open source DSP engine we could put between our streamer and DAC.
Of course, no doubt they’ll start selling $30k AI upsamplers which will get rave reviews and those of us in the DIY community will smirk and do the same with a Raspberry Pi and some Python code.