Voicing, in linguistics, refers to either the physical production of vibration by the vocal folds as part of articulation, or the potential phonological distinction this allows, i.e. the distinct difference between units such as [b] and [p] in many languages.
In most languages, sound units vary by voice, and very often this is distinctive, leading to minimal pairs such as 'bat' and 'pat' in English. Phonologically, this is a two-way distinction, but different degrees of phonetic voicing also allow further phonological contrasts in many languages, i.e. 'creaky voice' and 'breathy voice'. Phonetically, voice varies by position in the speech stream and from language to language; English [b d g], for example, are only fully voiced between segments such as vowels that are themselves fully voiced, while in Spanish, these sounds are fully voiced by default. These partially and fully voiced plosives act as physical productions of the phonemes /p t k/ in both languages.
Obstruent consonants are most typically distinguished by voice, among others. English distinguishes /p t k f θ s ʃ/ from /b d g v ð z ʒ/ by voicing alone, though this generalisation is somewhat oversimplistic. For example, in English, the length of segments appears to cue listeners into the voicing more than actual vocal fold vibration itself; the vowel in 'cat', for instance, is rather shorter than the one in 'cad', and the final [t] is also longer than the final [d] in many positions. This use of phonetic length corresponds to phonological voicing. The phonetic voicing itself is peripheral. In other cases, phonological voicing is only contrastive in certain positions; in German, for example, syllable- or word-final voiced obstruents are disallowed, and in English, the contrast between /ʃ/ and /ʒ/ is only productive word-medially, as in 'mission' and 'vision'; initial and final examples of voicing, such as 'genre' and 'beige', are loanwords from other languages and are often pronounced [ʤ], the first sound of 'jot'.