Phonetic transcription (also known as Phonetic script or Phonetic notation) is the visual representation of speech sounds (or phonetics) by means of symbols. The most common type of phonetic transcription uses a phonetic alphabet, such as the International Phonetic Alphabet.

The pronunciation of words in all languages changes over time. However, their written forms (orthography) are often not modified to take account of such changes, and do not accurately represent the pronunciation. Words borrowed from other languages may retain the spelling from the original language, which may have a different system of correspondences between written symbols and speech sounds. Pronunciation can also vary greatly among dialects of a language. Standard orthography in some languages, such as English and Tibetan, is often irregular and makes it difficult to predict pronunciation from spelling. For example, the words bough, tough, cough, though and through do not rhyme in English even though their spellings might suggest otherwise. Other languages, such as Spanish and Italian have a more consistent (but still imperfect) relationship between orthography and pronunciation. In contrast, a few languages may claim to have a fully phonemic spelling system (a phonemic orthography).

For most languages, phonetic transcription makes it possible to show pronunciation with something much nearer to a one-to-one relationship between sound and symbol than is possible with the language's orthography. Phonetic transcription allows one to step outside orthography, examine differences in pronunciation between dialects within a given language and identify changes in pronunciation that may take place over time.

A basic principle of phonetic transcription is that it should apply to all languages, and its symbols should denote the same phonetic properties, whatever the language being transcribed. It follows that a transcription devised for one individual language or group of languages is not a phonetic transcription but an orthography.

Phonetic transcription may be used to transcribe the phones of a language. In all transcription systems, there is a distinction between broad transcription and narrow transcription. Broad transcription indicates only the most noticeable phonetic features of an utterance, whereas narrow transcription encodes more information about the phonetic details of the allophones in the utterance. The difference between broad and narrow is a continuum, but the difference between phonemic and phonetic transcription is usually treated as a binary distinction. Phonemic transcription is a particularly broad transcription that disregards all allophonic differences (for example the differences between individual speakers or even whole dialects of the same language). Phonemic transcription provides a representation only of a language's abstract word-distinguishing units of sound (phonemes), and thus is not really a phonetic transcription at all (though at times it may coincide with one). Instead, a phonetic transcription focuses on more exact articulatory or acoustic details, whether more broadly or narrowly. A transcription which includes some allophonic detail but is still closely linked to the phonemic structure of an utterance is called an allophonic transcription.

The advantage of narrower transcription is that it can help learners to produce exactly the right sound and allows linguists to make detailed analyses of language variation. The disadvantage is that a narrow transcription is rarely representative of all dialects or speakers of a language. Most American, Canadian, and Australian speakers of English would pronounce the /t/ in the word little as a tap [ɾ] and the initial /l/ as a dark L (often represented as [ɫ]), but speakers in southern England pronounce the /t/ as [ʔ] (a glottal stop; see t-glottalization) and the second /l/ as a vowel resembling [o] (L-vocalization). Thus, on the one hand, phonetically, little can be represented as something like [ˈɫɪɾɫ̩] in many American, Canadian, and Australian accents but [ˈlɪʔo] in a southern English accent. Furthermore, in Australian accents especially, the first-syllable vowel of little tends to be higher than in North America, leading to the possibility of employing an even narrower phonetic transcription to indicate this, such as [ˈɫɪ̝ɾɫ̩]. On the other hand, a broad phonemic transcription of little is also possible that ignores all the above specifics of these aforementioned dialects; this can be useful in situations where minor details are not important to distinguish or where the emphasis is on overarching patterns. For example, one typical phonemic transcription for the word little is /ˈlɪtᵊl/, as is common in both British and American English dictionaries. (Slashes, rather than square brackets, are used to indicate phonemic rather than phonetic representations.)

A further disadvantage of narrow transcription is that it involves a large number of symbols and diacritics that may be unfamiliar to nonspecialists.^{[citation needed]} Broad transcription usually allows statements to be made which apply across accents and dialects, and is thus more appropriate for the pronunciation data in ordinary dictionaries, which may discuss phonetic details in the preface but rarely give them for each entry. Most linguists use a narrow transcription only when necessary, and at all other times use a broad transcription.

Most phonetic transcription is based on the assumption that linguistic sounds are segmentable into discrete units that can be represented by symbols. Many different types of transcription, or "notation", have been tried out: these may be divided into Alphabetic (which are based on the same principle as that which governs ordinary alphabetic writing, namely that of using one single simple symbol to represent each sound) and Analphabetic (notations which are not alphabetic) which represent each sound by a composite symbol made up of several signs put together.