I have been thinking about this issue and relating it to my experience in Photo Editing, I realise it sounds tenuous but bear with me, my business is in imaging training so I can see a parrallel.
If you take a an 8 bit image file and apply adjustments to it you get an issue called posterization, if the adjustments are great enough it looks terrible. If the same file is up sampled to 16 bit, it looks the same but when you apply adjustments due to the way the program reallocates data you get much smoother tonal transitions. Upsampling works really well but only if adjsutments are applied otherwise the file looks the same.
I imagine with sound the situaltion is the same up sampling will work really well but only if the adjsutments applied after up sampling are intelligent and sensible and that is I imagine where something may well be gained and I suspect lots of experimentation could produce some pretty amazing results beyond those currently available simply because no one size fits all solution is ever the best......every image file is different, and the same with sound recordings.
Adding a tube buffer would I think be like adding a low level noise filter to a image file, the effect is to apply very low level noise over the entire image (thought with a dvanced editing this is very adjsutable, but in the analogue domain of the tube buffer it would not be I imagine, basically all or nothing). The noise smooths the transitions between tones by adding extra tones in between, but it adds an overall grain effect that also removes micro detail and definition. To the naked eye at a distance it looks neat and film-like if done right, but the damage on close inspection is obvious. It does help compensate for poor files but it only really works well if you massively upsize/sample the file first and then apply the noise in the upsized format, before downsampling which will compress the blending a bit to add back a little punch.
Overall I think intelligent up-sampling with clever application of noise applied digitally to only limited parts of the audio spectrum will work to make harsh recordings acceptable, but it all has to be done in the digital domain.
My final idea then is this...take the recordings transfer them to a computer as 24 bit files, edit to suit using a suitable equipped editing program (I need to check which ones have the right tools) then re-burn as 24 bit DVD_V files and play then on a good universal disc player. This definitely works well with LPs I have dne is many times and I plan to digitise and treat my whole LP collection this way when I have a better soundcard and phono pre on the Mac.
I imagine the cost overall would be less than the tube buffer or any number of flash bits of gear too, and the really neat thing is you can build into the re-recorded music the exact EQ and ambience settings you need for your music listening room!
Have a great weekend- I'm off to build a circuit or two.
