The definitive translation in English is considered to be Andrew George's, which is available in a Penguin paperback version with just the text and a few notes, and a 2-volume critical publication by OUP. The two-volume work contains thorough notes on every reference in the text of the epic, and history of the source tablets, but a hard copy is very expensive. Project Guttenberg has older versions and are very out of date.
A few things need to be understood before you go looking for glosses or talk about 'definitive' translations, however. The epic of Gilgamesh comes down to us in a variety of formats. The story is originally Sumerian, but exists as part of the shared cultural identity of the Babylonians. The most complete forms of the text are in Akkadian (old, standard, or neo- Babylonian), with partial texts available in Sumerian and Assyrian. There are variations in dialect and in content, which is to be expected since it was one of the most reproduced of all Babylonian texts. The Andrew George translation is not based on the Sumerian version I linked you to above, but to the OB version in Akkadian, which you can find fully transliterated here.
Akkadian and Sumerian were unrelated languages whose speakers enjoyed a bilingual cultural interchange for centuries before Sumerian died out as a spoken language and later a literary one. In adopting the Sumerian writing system for Akkadian, Babylonians retained a substraum of Sumerian logograms, some fully adopted as loanwords and some merely condensed writing aids. The closest modern comparison is the use of alternately Japanese or Chinese 'readings' of specific kanji in the writing of Japanese. A truly complete exegesis of the Gilgamesh epic is not possible without a knowledge of both Sumerian and Akkadian.
Just to be clear, transliteration refers to a sign-by-sign representation of the original cuneiform text - this is not a pronunciation guide, and it will not help you look items up in dictionaries. It is, however, useful for finding specific signs in sign lists and co-ordinating with the original tablet. For example, [a-na LUGAL-ti] represents four signs on the tablet, delineated by hyphens, but it would be read ana šarrati(m) 'for the king', in NB Akkadian. LUGAL is the Sumerian word for king and replaces the root of the Akkadian word in writing as a logogram. At other times consonants or vowels are doubled, or tripled, in the adding of signs, an inelegance of the Akkadian syllabary; writing qabûm 'say, speak' could be rendered [qa-bu-um] or [qa-bu-u-um] or [qa-bu-u]. The point of all this is that an amateur will have a hard time working backward from a transliteration to get a word whose meaning can be searched and analyzed, and of course the corpuses you can search online do not typically provide glosses, a morphemic breakdown useful to language learners. To that end if you're enthusiastic about a study of these languages, I recommend the Huehnergard text A Grammar of Akkadian.
The Gilgamesh text itself is fragmented and pieced together from various sources written at various dates in different places, with newly discovered lines having been added to the corpus as they were found. This is a continuing practice; the most recent find was 20 additional lines belonging to Tablet V first identified in 2011. This paper includes the new section's material in a complete gloss of Tablet V. No published translation has yet included these fragments.
A simple transliteration and translation of the strictly Sumerian-language fragments of the text, to answer your original question, can be found online in any of the major corpus catalogs of Sumerian literature, such as The ETCSL. The Gilgamesh corpus is on this page sections 1.8.1.1 through 1.8.2.4, in transliteration and translation, although not side by side. It takes some getting used to, but you can use this resource to source each translated word to the original signs used to represent it.