DACs are often buffered. If you're transporting audio over Ethernet, the packet will contain some number of samples. Those samples have to be buffered. (For exa

DACs are often buffered. If you're transporting audio over Ethernet, the packet will contain some number of samples. Those samples have to be buffered. (For example, you might send 128 or 256 samples over Ethernet... obviously, the samples inside the packet are actually getting transmitted as a whole packet, buffered, and then sent to the DAC afterwards.) On top of that, you can synchronize the DAC clock to the clock recovered from the digital signal. For example, I routinely record data (and monitor) from an S/PDIF link. The standard way this works involves clock recovery with a PLL. That way, you only need a stable clock on the DAC--as long as the digital source remains within tolerances, it's fine. Even though S/PDIF transmits digital audio one sample at a time at a fixed rate, you don't need to worry much about jitter. We know it works this way because you can configure the gear to use internal or external synchronization. If you try recording S/PDIF with an internal clock, you get clicks and pops.