While I hope the time has led the questioner to the correct understanding of this problem, I find it a rather interesting question being asked.
It's true that a precise phonetic analysis reveals differences between realizations of the vowel //u// by its environments, or generally, that a vowel's sound quality slightly differs according to its preceding (succeeding) consonants in most languages of the world. But as a whole, differences inside a vowel don't grow as large as those between different vowels in a specific language or dialect.
Based on what as far as I can get from the question's way of thinking, and if I'm guessing correctly that mentioned "Chinese" is Standard Chinese, it's highly probable that OP was just misled.
First, there's no approximate Standard Chinese counterpart of Standard Japanese //u//. It's more backward than ü in SC lü, more fronted than u in SC lu, and more rounded than i in SC si (which is a wicked "vowel" that almost is a consonant in phonetic perspective, but ironically being closer to what SJ //u// sounds especially when devoiced).
Second, not all consonants adopt aforementioned vowels in SC. The ü only appears after n, l, j, q, x, and the -i after z, c, s. Only u gets along with any consonant.
Therefore, if you have Chinese ear (it's sort of mindset rather than hearing...) and hear SJ //su//, you can translate it to closest SC si, but when you hear SJ //ku//, you can't make you believe it's ki (sorry for abusing pinyin) or kü because they're not in your phonological inventory, as the result you think it must be ku.
Generally speaking, there's little chance that a sound of a foreign language hit the "sweet spot" of any sound you already know. Whatever it is recognized to you, I bet it must be more or less "off" from decent sound in your native tongue. Knowing to accept those "waffling" sound as is would be the first step to learn foreign pronunciation.