This is the reason people use IPA (International Phonetic Alphabet) in the first place.
You cannot simply guess exact pronunciations by orthography! Especially the English language!
For example, to address Kiam's first point of, say, the asker's first language lacks an /h/ (e.g. Romance languages). English orthography is so irregular that sometimes, the H is rightfully silent (like honour, hour, ...), while in other cases it isn't! (high, hello, hat, hose, human, hacker, ...)
And for another case of orthography, different languages read different letters differently, as demonstrated by the /h/ example above. J, ch, and E? Don't get me started.
These sorts of "how to produce a foreign sound present in English" questions must be answered from scratch. We must tell these people the specific respiratory movements in order to articulate these sounds in the first place.
The problem can go in reverse too. English lacks a /x/, present in several languages, frequently transforming it into /h/ or if word-finally or Greek, /k/. A way to teach this is to start with /k/ and instead of doing a stop movement, do a fricative movement.
General "how to pronounce X" can be answered with IPA, and link to a key on how to pronounce the IPA symbols, like to Wikipedia (which tells the specifics on the articulatory movements).
Or, to make it easier to comprehend, if the missing sound is close to a sound the asker already knows, tell him to start with the close sound and then specifically how to work your way towards the missing sound. This is probably the most preferred, if possible. But say, the asker's first language lacks a whole class of sounds (e.g. English lacks uvulars, which are present in French, and a ton of languages lack schwas, so on), we're in a whole load of more trouble.