Hmm, not sure about that, it seems to me it just stores the audio in one series of small fragments, which are just split up somehow by time.
Did you think about recording the audio with something?
Like OBS or Audacity (versions before 3). I think that may be a semi-good solution. Or, for somewhat better quality you could download the whole track (yt-dlp can handle it), and then cut it for the pieces you want to keep, in Audacity (before version 3) or something else