Sources include:
John Wells's page on Phonetic Symbols for English (Wells)
Cambridge English Pronouncing Dictionary (EPD)
Longman Pronouncing Dictionary (LPD)
Oxford Advanced Learner's Dictionary (OALD)
Cambridge Advanced Learner's Dictionary (CALD)
Collins Cobuild Advanced Learner's Dictionary (CCALD)

While the sources are unanimous in their use of the standard symbols, they differ in their treatment of the additional symbols. Where possible, the various different forms are given together with an indication of which source uses which alternative.
 

There are two L sounds in English: a 'clear L' and a 'dark L' (or velarised L). The difference is produced by the position of the back of the tongue which is raised for the latter.
 

The clear L is symbolised by /l/ and the dark L by /ɫ/.

Clear L preceeds a vowel or /j/.

Dark L is followed by silence or a consonant other than /j/.
 

Despite the fact that they are strictly speaking not IPA symbols, /ɪ/ and /ʊ/ are used by all the sources to denote two different things.

1) They denote a middle -i- or -u-, that is longer than /ɪ/ or /ʊ/ but shorter than /iː/ or /uː/.

2) They denote denote variable pronunciation where a vowel could be pronounced as short /ɪ/ or /ʊ/, mid-length /i/ or /u/ -as above- or long /iː/ or /uː/.

 
There is no easy way to know which sense the symbol is used in. But in practice this does not present a problem, at least for Italians, who tend to pronounce such vowels /i/ or /u/, as in Italian.

In fact /i/ or /u/ are close to the standard Italian vowels -i- and -u- but in English they are only found in unstressed syllables.
 

The OALD illustrates 3 finer points:
 

i) When /i/ is followed by /ə/, the combination can be pronounced /ɪə/, /jə/ or /iː.ə/.

ii) Similarly, when /u/ is followed by a consonant, it can also be pronounced /ə/, giving at least three possible pronunciations /ʊ/, /ə/ or /uː/.

iii) Also, when /u/ is followed by /ə/ the combination can additionally be pronounced /wə/.

In BrE r is pronounced or mute as follows:

1) Before a vowel or diphthong, r is always pronounced, and is transcribed /r/.

2) Before a consonant, r is always mute and is not transcribed.

3) At the end of a word, r is pronounced only when the following word starts with a vowel or diphthong. This r is called potential R and is transcribed /^r/(EPD, CALD, CCALD), /^ər/ (LPD) or /(r)/ (OALD).

In AmE r is always pronounced, either as a normal consonant or as part of a retroflex vowel. American treatment of r is dealt with under British and American English below.
 

In effect the mute and/or potential r is used to 'write' several vowels and diphthongs.

It is found in many spellings of schwa /ə/ and is frequent in the long vowels /ɑː ɔː ɜː/. All the standard spellings of the diphthongs /ɪə eə ʊə/ have an r.

WHEN POTENTIAL R IS PRONOUNCED

potential R is reinstated in two cases:

1) When one of these vowels or diphthongs finishes a syllable, and when the following syllable starts with a vowel, the r is reinstated at the beginning of that syllable*.

2) When one of these vowels or diphthongs finishes a word, and when the following word starts with a vowel, the r is reinstated.

This process is called linking, and the r 'effectively' attaches to the beginning of the following word.

* The sources all give /ˈtʊər.ɪst/ etc, with the r at the end of the stem, but it is useful for language learners to perceive this process as analogous to the linking described in 2.

Learner's dictionaries do not have to give variant pronunciations, and doing so can create more confusion than certainty, except when a particular variant is very common.

Instead, specialist pronunciation dictionaries must try to track as many non-dialect pronunciations as possible. Also, they must observe more subtle sound differences, which are probably imperceptible to learners.
 

3) An r only pronounced in determined conditions is represented as ^r or (r).

See potential R for details.
 

4) Only in EPD and LPD.

Note that sounds lost through elision are not treated in this way.
 

The consonants l, n, m and r sometimes form syllables with no vowel. They usually do this alone but can also combine with another consonant. These are called syllabic consonants and are denoted by a small dot under the phonetic symbol /l̩ n̩/.

But some people pronounce some of these syllabic consonants as a weak syllable with schwa /əl ən əm ər/. When both pronunciations are possible a superscript schwa /^əl ^ən ^əm ^ər/is used to indicated choice.

Thus, in middle the syllabic consonant is standard: /ˈmɪd.l̩/.
In arial only the syllable with schwa is standard: /ˈeə.ri.əl/.
But in most cases both models can be used, as in classical where /ˈklæs.ɪ.k^əl/ means either /ˈklæs.ɪ.kl̩/ or /ˈklæs.ɪ.kəl/.

The difference is small so the sources differ in notation: Wells uses /l̩/ and /n̩/ while EPD and CALD only use /l̩/, while LPD and CCALD use niether. OALD differs from the rest in using neither the superscript schwa nor a syllable separator.
 

Syllabic L is found in most -le -ible and -able endings and is denoted by:
/l̩/: Wells, EPD, CALD, eg /ˈmɪd.l̩/
/^əl/: LPD, CCALD, eg /ˈmɪd.^əl/
/l/: OALD, eg /ˈmɪdl/

Instead, according to EPD, -el -al and especially -ful endings are usually variable /-^əl/.
 

Syllabic N, as denoted by /n̩/, is only used by Wells and his examples are:

Endings in -an -on and -en are usually /-^ən/.
 

The suffix -ism and other endings in -sm are pronounced /-.z^əm/ or sometimes /-z.^əm/.
 

Some central syllables ending in r are pronounced /-^ər/.
 

Stress, also called lexical stress, word stress or word accent, is the emphasis of a complete syllable by pronouncing it more loudly and openly. It is different from accent which indicates a particular pronunciation of a vowel.

Primary /ˈ/ and secondary /ˌ/ stress marks are used by all the sources immediately before the stressed syllable. Stress is not indicated on monosyllabic words. Secondary stress is only indicated where a second syllable is stressed above the unstressed syllables. Stress marks replace the syllable marker before the stressed syllable.
Examples: fəˈtɒ.grə.fi   ˌfəʊ.təˈgrəf.ɪk   ˌfəʊ.təˈgrəf.ɪk.^əl.i

Note that some textbooks prefer to use a system of large and small spots above a word to indicte stress and syllables.
 

Division between syllables is indicated by a dot /./ in all the sources except OALD, which marks syllable division in the headword, and LPD, which uses a space. The syllable marker is not used in conjunction with a stress mark.
 

Knowing where syllable division occurs is especially important where composite consonants /tʃ/ and /dʒ/ are found, since - in some cases - the individual sounds exist in separate syllables, and must be pronounced separately.
 

A diphthong is a sound that starts with one vowel sound and finishes with another, a triphthong is a sound that moves between three vowel sounds in the same syllable.

In practice triphthong describes a combination of one of the diphthongs /eɪ əʊ aɪ aʊ ɔɪ/ plus schwa, when the schwa is considered not to have sufficient force to form a new syllable.

Experts disagree about whether triphthongs exist in a technical sense, but native speakers treat these sounds as if they were unified into one syllable so most authorities agree that the concept is useful.
 

The length mark /ː/ indicates a long vowel. There are 5 long vowels in English: /ɑː ɜː iː ɔː uː/. It is also used in 3 of the 4 French-style nasal vowels: /æ̃ ɑ̃ː ɔ̃ː ɜ̃ː/.

NB. Some versions of Internet Explorer have a bug which causes this symbol not to be visualised. If you see a square box (like this
) instead of two dots (like these ː), you should upgrade your browser.
 

The glottal stop is a forced pause made instead of a t. It is produced by closing the epiglottis, hence its name. It is represented by /ʔ/.

It is not used in standard English but is common in regional varieties of both BrE and AmE.
 

1) an obstruent
ie /p b t d k g tʃ dʒ f v θ ð s z h/

2) a nasal
ie /m n ŋ/

3) a semi-vowel or non-syllabic L
ie /w j/ or /l/

* Read the page from LPD which these notes are based on.

Modern British dictionaries give British and American pronunciations for all headwords. Differences can usually be predicted according to the following rules.
 

Vowels preceeding a potential or mute R in BrE are 'coloured' (technically: "influenced by rhotic colouring". EPD) by the r in AmE.

Note that /ɚ/ and /ɝː/ are called retroflex vowels.

EPD, CALD AND LPD use these symbols (as well as /t̬/, see below).
OALD AND CCALD simplify into /ər/ and / ɜːr/.

Here we are only interested in sounds from Scots and Welsh which most English speakers produce when pronouncing Scots or Welsh words or names.
 

The voicless velar fricative /x/ is used by English speakers in a handful of Scots words and in many names and placenames of Gaelic origin.

It is also used when pronouncing some common German words with -ch-, Spanish words with -j-, and a variety of other non-English names and placenames.

Note that /x/ is never used in modern English for -gh-.
 

EPD uses /hl/ (with an optional /h/, see optional sounds below) in order to represent this range concisely.

The 3 Advanced Learner's Dictionaries do not use either rendering (preferring /l/).
 

A large amount of English vocabulary comes from French. Almost all such words have been assimilated into English and anglicised in pronunciation and spelling. A few words, however, have kept their 'Frenchness'. These are pronounced in imitation of the original. Thus we are not concerned with correct French pronunciation, but with representing the sounds the British make when they try to approximate this.

Note that original French pronunciation is given here for comparison only.
 

The symbol /a/ is only used in OALD, and corresponds to a short open a, different to the standard English /æ/.

It exists in Italian, and so can help differentiate between the -a- in the two languages, which is useful in correcting the practice of pronouncing an /e/ in words like back and cracker.

There are four French nasal vowels /ɛ̃ ɑ̃ ɔ̃ œ̃/, which are reproduced in French words in varying degrees of accuracy by English speakers.

By convention these vowels are represented by their equivalent English vowel symbol plus a nasal glyph /æ̃ ɑ̃ː ɔ̃ː ɜ̃ː/.

Note that with the latter three vowels, this means that the English approximation contains a length mark /ː/ and two have a different symbol altogether.

The sources differ widely in both their adoption and their use of these symbols.
 

The nasal glyph is a squiggle /  ̴/ above a vowel symbol. It indicates that the vowel is pronounced nasally, usually forming an approximation of a French vowel, see above.
 

The use of additional symbols gives greater accuracy and flexibility to phonetic script, but it does so at the expense of immediacy and legibility.

Witness the difference between /prəˌnʌnt.siˈeɪʃ^ən/ in EPD and /prəˌnʌnsiˈeɪʃn/ in OALD.

For most language learners most of the time, immediacy is more important, but the application of a particular symbol can make a particular point clear.

Consequently I have adopted a varying use of additional symbols, keeping them in reserve to be used only when their presence will help a particular explanation.

For example, I have omitted the syllable marker /./ and potential R /^r/ from transcriptions of examples of the basic symbols, but have included both them and others in the examples of additional symbols and in the notes.

Because this might seem - at first sight - inconsistent, I have set out below how the additional symbols are employed on this site.
 

/əl/ is used in place of /^əl/ and /l̩/, except in discussion of syllabic consonants.

/ən/, /əm/ and /ər/ are used in the same way.

/^r/ is used in definitions, in examples requiring connected speech and in explanations on how and when to pronounce r.

It is not used in situations where the r is not pronounced, including in lists of words.

/(r)/ is not used on this site.

It is sometimes useful to be able to transcribe the IPA using common keyboard symbols. SAMPA is a system for doing this.

COMPLETE PRINTABLE LIST

SAMPA uses the same basic symbols as the IPA where possible. Below is a summary of those symbols which differ:
 

Several people have offered their voices for recordings:

David Nicholson: standard British English & glottal stop
to be decided: standard American English
to be decided: French
to be decided: Welsh