The original Finnish orthography as developed in the 16th century was unsystematic, and it imitated some of the oddities of Swedish and German orthography. It was later reformed and became rather systematic, mostly with one-to-one corresponding between letters and sounds. After this, the pronunciation has not changed substantially, so the simple correspondence has been preserved. This is good news to anyone who wants to learn Finnish, as well as to speech synthesis and speech recognition.
The correspondence is, to be exact, between letters and phonemes. As mentioned in chapter The phonemes in Finnish, a phoneme is an abstract concept in linguistics. For example, the “t” sounds in English “hat” and “tin” can differ from each other, but they are still recognized as the same; pronouncing the “t” in “tin” as normally in “hat” might sound a bit odd (the “t” sounds too weak, so to say), but it would not change the meaning of the word. Changing the “t” sound in “tin” to the English “th” sound would change the meaning. In Finnish, too, phonemes vary considerably, especially the “h” phoneme. For example, the letter h in the words hei, paha, and lyhty are pronounced very differently, but these sounds still represent the same phoneme. The differences are determined by the phonetic context, so they cannot make a difference in meaning.
There are actually several deviations from the one-to-one correspondence between letters and sounds, although they are small as compared with e.g. the writing system of English. The deviations, such as boundary gemination, are discussed in in different sections of this book and summarized in section Deviations from the phoneme principle: a summary.
Some of the sounds in Finnish are difficult, however. This does not mean any particular physical difficulty in producing them. Rather, it is a matter of sounds and phonetic contexts of sounds that are unfamiliar e.g. to many people who speak English as their native language.
A language learner needs to learn to pronounce unfamiliar sounds such as y and ö, to pronounce Finnish diphthongs, and to make a distinction between short and long vowels (even in unstressed syllables) and between single and double consonants. Some exercise is needed to say taka, takka, takaa, taakkaa etc. so that the listener knows which word you mean.
Although unstressed vowels can be somewhat shorter and slightly less clear than stressed vowels, there is no vowel reduction comparable to, say, how the first vowel of the word “about” is pronounced in English. Word-final vowels are full vowels, too; the three vowels of the word vasara (hammer) are basically identical, except that the first one is stressed. Similarly, the vowels in the word vene (boat) are identical.
In free speech, some final vowels are dropped, e.g. yksi (one) often becomes yks. However, there is usually no partial reduction: either the vowel is fully pronounced, or it is omitted. Partial reduction may appear in fast speech and in dialects. Dialects also have reduced (very short) vowels that may appear between consonants to ease pronunciation. For example, in Eastern dialects, you may hear kolome, with second o very short, instead of the standard kolme (three).
When you can handle the issues described above, you can pronounce Finnish quite understandably. Some additional effort is needed to produce more native-like pronunciation. For example, it is a basic requirement to pronounce the letter r audibly in all contexts, but the exact quality of this sound is less important for understandability. To produce a native Finnish r, you need to pronounce it as a fairly strong trill, as in Scottish pronunciation of English.