Method 2: Tiptoe Around the Problem
Here's what you do: Don't use one repeating tune for an entire level of a game. It seems normal and intelligent to have a "first level" tune and a "cave level" tune and an "underwater" tune, but it's just plain old school. There's no excuse, and it will badly injure the ears of anybody playing the game and kill anybody listening who isn't playing the game. So don't do it. Okay? Just don't. If any one tune in your game repeats for more than five minutes, you should do one of the following:
Change to another tune after five minutes.
Stick a hot fork into your own eye, you evil moron.
Reuse your resources in different circumstances. I know you want special "cinematic" pieces, and "payoffs," and a unique piece for the puzzle with the cute duckies, and such. But the math is simple. If the game's budget is for 20 minutes of music, and the game is constructed so that music plays for an hour in a given session, the music is going to repeat somewhat. And remember that three repetitions of the music would happen only in the best possible circumstances, meaning all music has the same odds of repeating. But suppose you get greedy about special-case music. The more of your music that goes to special one-time cases, the more the other tunes have to repeat to cover for it. Reuse that "Binky meets the cougar" tune as a "tense puzzle-building" or "will we win the pony race?" background piece. The players won't mind; the situation will be different enough that they'll experience it as two different pieces of entertainment. The people listening and not playing will be grateful for one less repetition of that incessant "riding the pony" music.
Do not use musical structures that utilize repetition to build familiarity. This is hard to get away from. Sure, conventional musical theory suggests that we play familiarity against variation to achieve tension. That's why conventional music uses forms such as AABA. But in a game, you're going to get 30 repetitions of the tune at least. Think about that. How many times have you listened to the CDs in your house? Even your favorite CD? In a game, you can concentrate on the variation and relax on the repetition. An hour into the game, the familiarity will be there, I guarantee it.
Fade to silence after two minutes of inactivity. Some games don't, and I have one thing to say about that. It's a strong indication that everybody on the development team lives alone.
Positives: These are brilliant ideas, and I've seen more short-run success come from them than from any other ideas. The costs are low, and the results are good. Do the above, and if your audio content is listenable and your game is playable enough to ship, you will put your game into the top five percent of great-sounding games.
Negatives: You will only be in the top five percent of great sounding games, and with all respect for my chosen profession, that's not yet something that everybody would want to subject his worst enemy to. Because why? Because this: Because even after you've executed all of this earnest cleverness, you're still trying to make an hour of music palatable for 40 hours. Frankly, a miracle is wanted.
Miracle, you say? Surely, Fat Man, that is too strong a word, and you are exaggerating the extent of this problem.
Perhaps you're right. Miracle is a strong word. And all we have to do is make an hour of audio suffice for 40 hours, which isn't that tough. It'd just be like making a loaf of bread suffice for 40. No miracle needed. No need to use such strong, potentially offensive language like "miracle." Let's try another word.
So instead, I'll put it this way. Making one hour of music suffice for forty hours of gameplay is like polishing a turd, and you can't polish a turd. Well, you can, but it's difficult, it gets your rags all dirty and smelly, and in the end, what do wind up with? A shiny turd. And you can't feed the masses with that. They prefer bread.