I know you simply mistyped (by writing “unvoiced” twice), but I can’t figure out what you intended to say here. Can you please clarify?
In English, the letters <t> and <d> usually represent the phonemes /t/ and /d/, as in the words “tip” and “dip”. At first inspection these appear to be pronounced identically except for voicing—that is, in both cases you put the end of your tongue on your alveolar ridge and quickly release it, but with /d/ you also engage your vocal cords. However, that’s not really the whole story—in English /t/, like many other unvoiced consonants, is aspirated, meaning that you release a puff of air when you make the sound. But voiced consonants, like /d/, are not normally aspirated in English. You can check this yourself by putting your palm in front of your lips and speaking the words “tip” and “dip”. You’ll feel a much stronger puff of air for the first word.
Whispering is basically removing the voicing from all sounds, so while whispered /t/ and /d/ sound almost the same, if you listen carefully you might still be able to differentiate between them by listening for the aspiration. Aspiration isn’t phonemic in English, which means it doesn’t serve to distinguish one sound from another in normal speech, but whispered speech is one unusual case in which it actually can.
-1, especially for that “Purple Monkey Dishwasher” remark.
This post has been purpled by the purple monkey dishwasher!
When I was a kid ('60’s U.S. Midwest) it was called Chinese Telephone.
Calling it now – **ascenray **is going to come into this thread and let us know that this is not, in fact, true of Hindi 
And I believe he’ll be right about that.
Every human language has more “signal” than is strictly necessary to promote communication. That’s why accented speech, disordered speech, baby talk, and such can be understood at all (with varying degrees of difficulty). There are even more clues than those apparent in the sonic ascpects of speech, making lip reading possible.
…
Psychonaut, I’d actually expeect Rotokas to suffer a lot less when whispered than typical languages due to that selfsame tiny consonant inventory.
How so? English has 24 consonants, of which there are 8 voiced–unvoiced pairs. So ignoring the aspiration issue, 67% of the whispered consonants are potentially ambiguous. Rotokas has three voiced–unvoiced consonant pairs, and no other consonants in its inventory, so 100% of the consonants have the potential to be ambiguous when whispered.
The only way you could claim it’s less ambiguous is if its phonotactics disallow certain consonant–vowel combinations in such a way as to resolve the ambiguity. (For example, if the syllable /bi/ is permitted but /pi/ is not, then there’s nothing ambiguous about whispering /bi/.) However, this doesn’t appear to be the case: Rotokas syllables are permitted to be V, VV, CV, and CVV, and the syllable inventory size is 350; if you do the math (for 6 consonants and 10 vowels) you’ll see that exactly half the possible combinations are indeed permitted, which is extremely permissive as far as languages go.
Judging from bordeland’s post, I’m guessing he/she will reply that any language with such a small consonant inventory is probably “making up for it” by having an even more than usual inventory of non-phone-related ways of expressing words – that is, all the other redundant stuff we (subconsciously) do keep the signal clear through all the various kinds of noise.