Sorry, but we are not here to do your college project for you. Start by thinking about what signal breathing would look like in audio terms. That means, you would start with an audio sample that has noticible breathing in it. Plot what you have. Think about it. Is there a periodiic behavior that would indicate breathing? Can you find some way to recognize that characteristic signal? What tools do you know about for such a purpose? Look at several audio tracks that have breathing. Is there some similar characteristic among them? Next, look at some audio that has no recognizable breathing in it, but stil someone talking. Are there different characteristics?
What do you know about signal processing? Do you have skills in machine based pattern recognition? Is it possible that some sort of deep learning approach might work?
Again, this is your project, not ours. You are the one who gets the grade, based on you thinking about the problem, and learning to use tools that will help you. We get no credit. Surely if someone tells you exactly how to do the task, your project won't won't have their name directly next to yours, will it? Would we get a degree for doing your work?
Do some reading online. I'd bet you can find lots of stuff published. (I know that to be true, since I just did a quick search.) GOOGLE IS YOUR FRIEND.
I might try a search using terms like: "detection of breathing in audio", which might be a good start. Or perhaps "human breath detection microphone".
