The only way to determine the exact nature of your "l" is with a broader corpus of recordings plus some actual physiological data. L is subject to various degrees of backing and vocalization not only in English but many other languages. In my dialect of American English, /l/ is retracted, but still lingually articulated, when preceded by a back vowel in the same syllable (however: /l/ is completely deletes after /ɑ/ in certain clusters, e.g. wa(l)k, pa(l)m where some people retain l). In some dialects, this backing also applies in milk, help, fill. It resembles [w] to the point that you might well transcribe milk as [mɪwk]. It is thus possible that in your dialect, l has simply become w in some coda context. The question then is whether "code" and "cold" are pronounced the same or different for you. You also want to look at l intervocalically, as in "folly, allow", because stress affects syllabification which may affect how l is realized.
There is no right or wrong in terms of pronunciation, just confusing or not, from a TESOL student's perspective. There is no one standard, rather there are many standards, for example US standard, RP for British Standard, Kenyan Standard, etc. If you're aiming for RP, sorry, you missed the mark, but it sounds pretty okay for SA standard.
The downward formant transition at the end of the vowel in in your "golf" clearly indicates that there is a w-like element for "l". In the case of "also", there is nothing at all. I suspect that the vowel quality is different, and different from "ostrich", "awesome" or "arse" (if you have the arse/ass distinction).