There is a difference between upsampling and oversampling, When ratio is
even then it is oversampling, usually achieved by usage of PLL. It reduces
jitter - that translates to noise in frequency domain. Stronger reduction of
jitter can be achieved by up sampling in asynchronous rate converter. It can
be done in the DAC or in the re-clocker. Ratios are not even anymore and
often not published. My Benchmark DAC1 upsamples to equivalent of
about 1 million times, but outputs data only at 110kHz where D/A chip THD
are lower than THD at 192kHz. Sound free of noise is often called sterile,
clinical etc. My first impression of DAC1 was that some instruments have to
be missing - it sounded too clean. It was also purposely designed not to
sound warm. It can sound bad in some systems (sounded bright with
previous speakers) but it sounds wonderful with warm sounding speakers.
Any form of filtering automatically increases resolution. It is interesting that
opponents of oversampling and Delta-Sigma DACs like the sound of SACD
that uses exactly the same principal. There are great DACs in each
category, as Bombawalla said, not to mention that it is very personal. Clean
sound can be sterile, while adding a little bit of noise makes sound more
vivid.