Intermission: Consonant irregularity
The italic text in this intermission will indicate values in the International Phonetic Alphabet.
The consonant g is sometimes pronounced ng, even at the beginning of words. In modern times, this is a matter of accent, not variation between dialects. It feels prestigious, not slangy, but it's being phased out.
The syllable ん is always pronounced as m before "p" and "b", always pronounced as ng (ŋ) before "k" and "g", and always pronounced like ng (ɴ) at the end of an utterance. Basically, it merges with whatever comes after it, just like how the "input" is usually pronounced "imput", and "inking" as "ingking".
Most japanese syllables use the consonants regularly. Only a couple don't.
Within the basic kana, only the kana ふ, つ, し, and ち have irregular consonant sounds. They are, in effect, "fu" (ɸu), "tsu", "shi" (ɕi), and "chi" (t͡ɕi), in order.
When they're voiced, three of the four irregular basic kana, づ, じ, and ぢ, change qualities slightly. So does ず.
In standard japanese, ず and づ are in free variation. Either one can be pronounced as dzu or zu.
The same is true with じ and ぢ pronounced as ji (ʑi) (as in the second consonant in "vision") or dji (d͡ʑi) (as in the first consonant in "James").
ず・づ and じ・ぢ being treated the same way is a feature of the standard accent. Other accents can merge the four of them together even more, or not merge them at all.
When し, ち, じ, and ぢ start diphthongs, like ちゃ, ぢゃ, しゃ, and じゃ, the "i/y" sound in the middle is dropped. For example, ちゃ is always pronounced as "cha" (t͡ɕa), never as "chya" (t͡ɕja). Normal diphthongs like きゃ are pronounced the normal way, like "kya" (kja).