Some of the article worried me a bit. I think I know what's being said, but I thought it was fairly easy to misunderstand things. One of those things was reporting WAV as "100% compression ratio" ... in my book, WAV has no compression, so is 0%. I was also uncomfortable about Samples/mcs ... I presume (never presume!) that mcs is microseconds. I worried that I was filling in too many blanks for myself. Of course any compression is going to handle data as blocks, not samples, so perceived latency will be block-size dependent.

An old PC will likely have a mechanical hard drive and the biggest impact then on random playback, assuming the data is still to be read, will be seek and read time. If all the data is already read in, then it's fairly likely that its already in decompressed for inside the application.

In an embedded sampler with which I was briefly involved, samples were requested from bulk storage at the moment of a MIDI note. The software then came back 'later' to collect the new instrument and if the instrument wasn't ready, it just waited for it, letting the existing sound continue to play instead. How much, I wonder, would one actually notice a, say, 10ms delay before a new sound started? The impression I had was that the delay was very rarely anything like that long, though the data was in SSD.

If one were to start play at a random point, would one even notice a 50ms delay before the sound started?

Jazz relative beginner, starting at a much older age than was helpful.
AVL:MXE Linux; Windows 11
BIAB2025 Audiophile, a bunch of other software.
Kawai MP6, Ui24R, Focusrite Saffire Pro40 and Scarletts
.