Scientific hypotheses are not drawn out of the air, but are directly suggested by examining the structural elements of the situation that produced the problem. For example, if your problem is that you want to find a way to permit people to communicate over long distances without shouting, then you will be thinking specifically about ways to modify the voice or amplify the ear; and your hypotheses will reflect the possibilities you envision.

Exactly how you go about envisioning productive possibilities is, unfortunately not something one can spell out in a recipe. It frequently requires a kind of genius that permits you to see analogies between what you know of the problem and what you know of the world. And indeed this is what Alexander Graham Bell apparently did in inventing the telephone:

It struck me that the bones of the human ear were very massive indeed, as compared with the delicate thin membrane that operated them, and the thought occurred that if a membrane so delicate could move bones relatively so massive, why should not a thicker and stouter piece of membrane move my piece of steel.

Clearly, we touch the tip of a very big iceberg here. No one knows what makes an apt analogy occur to one person and not to another. Certainly having total knowledge of the problem situation helps, as does spelling out and re-examining all your assumptions about it. What we do know from those who have written about the process, however is that their insight when arrived at is always a visual image.