I do not think his presentation about jitter is correct. This is from that article.
View attachment 152026
This seems to imply that some clock jitter on the wire in the data corresponds to DAC jitter that results in AC waveform distortion. I do not believe that is true. Some DAC chips with built-in USB do try to recover the audio clock from the embedded signal, but that is really an issue with the device's clock. I thought that all good DACs use their own high-quality clock isolated from the I2S or USB clock.
This is from the USB 2.0 standard, page 67, when discussing isochronous transfers (I added the bold):
USB is not simple TTL signaling, like that article shows. It is a NRZI encoding of multiple voltage levels that yield different eye diagrams depending on the USB version.
The article is about USBC (usb 4). USB 4 is nothing like early USB. It is Reed-Solomon FEC encoded, which means the bits are scrambled. USB 4 interfaces have re-timers in them between the cable and a DC-blocking capacitive coupler and the host buffer. It uses a spread-spectrum signal. I believe all the services, like isochronous transfer, are handled by upper layers. They no longer exist on the wire as separate signaling.
I'm not a USB implementor, so maybe I'm missing something here.
Marc