Also, for the non Dutch speakers amongst us, in the first couple of minutes the guy talkes about how adding lowpass filters to get rid of the images caused by nyquist are bad because they cause pre and post ringing artefacts on impulse type signals, and proposes that making use of the natural frequency roll off of the entire signal chain including the human ear and the speakers will accomplish the same goal without these issues. He also mentions a whitepaper, of which an English version resides here :
http://www.metrum-acoustics.com/Design%20Philosophy%20Metrum%20Acoustics.pdfI for one wonder what is going to be said about intermodulation distortions etc, somehow this idea intuitively seems a bit "off", but its interesting nonetheless.
Edit: Watched the whole thing now, not a word was mentioned about IMD etc, but here follows a quick summary.
First they talk about how dithering gets a 16 bit CD from 96 to 120db of dynamic range.
They made their own dac implemented with 2 12 bit r2r ladders and an fpga, both dacs are summed in the analog domain to supress the noise of the switches in the dac, this gets them to 20 bits. They needed to roll their own chips since other non-oversampling chips were end-of-life
Next up they refer to Nelson Pass's research that states "faster circuits sound better", meaning higher bandwidth. Their resistor ladder does 50Mhz sampling rate if they clock the data in parallel, but using the built in shift register is slower, since 44.1K is good enough for audio anyway. It is mentioned that SACDs do sound a bit better because 24Bit 96K, but that its mostly the better mastering process.
Other manufacturers can buy their DAC modules, you only need to stuff in i2s and power to get out audio.
The next hurdle they have to overcome is to make the 20bit dac module more affordable and
bring the price down from 5K to 1K. Its mentioned in passing that a lot of info about DACs on the web is misleading, and 24 bit converters are mostly a marketing driven thing, but they think some barely discernible difference is still audible.
The last interesting technical bit is about the noise floor of single ended versus symmetrical connections. The single ended scope plot shows large distortion at -90dbFS, whilst the symmetrical signal still looks fine until -120dbFS. The distortion products on the single ended connection are basicallly 50Hz mains and its harmonics.
The video ends with a short factory tour, standard affair, p&p and reflow.
Ok, I hope that whole story gives the the non-dutch speakers a bit of insight into what is going on in the video. Although there is no quackery going on, and most of the experiments done make total sense, relatively little time is given on how leaving out the anti-aliasing filters on the DAC does not cause any issues of its own, which is a bit disappointing.