I had never come across the concept of an unvoiced vowel, never mind an unvoiced schwa, but it turns out that it is possible to record such a phenomenon with a spectrogram.
Norman Lass in Contemporary Issues in Experimental Phonetics gives the word commercial
as an example: "The /k/ sound appears to be unusually long because of the unvoiced quality of /ə/". Figure 6.18 in this reference is particularly illustrative.
Based on this reference, I would expect the same phenomenon is observedEDIT
@StoneyB explains in cut
the comments that the unvoiced schwa does not happen with stressed syllables and thus it doesn't happen in subbut
, but I wouldn't expect it is the case for butcut
becauseor bsub
is a voiced consonant.