In the audio world, “generation” is a term from the days of recording on analog tape. With analog tape there are a limited number of tracks available (2, 4, 8, 16, or 24, depending on the format). Engineers were often forced to mix or merge sound from two or more tracks down to another track on the same reel in order to free up the first tracks for recording new vocal or instrument parts. The process went something like this: Record audio to tracks 1-6 on an 8-track tape machine. Mix those tracks down and record the result to track 8. Erase tracks 1-6, and record new audio on those tracks, effectively giving you 12 tracks of audio (the original tracks 1-6, plus the new tracks 1-6). Those tracks could be mixed down to track 7, and new audio recorded on 1-6 again. At some point, tracks 7 and 8 might be mixed and recorded to another track, etc., etc., etc. Many early multitrack recordings — such as classics from The Beatles and many others — used this technique.
Each subsequent copy of an original audio source is known as a “generation.” So the original recorded audio is “first generation,” and if that audio is mixed to another track, the copy of the original audio on that track is “second generation,” and so on. The further you go from the first generation, the more generational loss the signal will experience.