In principle, yes. But it will only be easy if the words are spoken in isolation with an effortful clarity.
But in natural speech there is a high degree of variation in vowel pronunciation: both random variation and systematic variation between speakers and linguistic contexts. You might be able to settle the question you are posing conclusively if you already had a large corpus of transcribed speech for the speaker in question, so that we could assign a probability of the word being effect vs. affect.
In practice, this is completely impractical, and the thing to do is assume that the speaker is using the prescribed form here, affect. But for what its worth, I am hearing the speaker say [əˈfɛkt], which is how I pronounce affect. I would have expected a higher vowel, but still centralized (i.e.,
[ɨˈfɛkt]) if she was saying effect.