Yeah this has happened to me a couple of times and it genuinely threw me off. Was doing a session in my back room about three months ago, asked out loud if anyone was present and got back what sounded like "I'm here" on the recording. Not just noise, an actual phrase that matched the question.
The sceptic in me keeps saying its just audio pareidolia, brain finding patterns that aren't really there. But I've had two other people listen without telling them what I asked first, and both picked out the same words independently.
What recorder are you using? I'm on a basic Olympus digital one, nothing fancy. Wondering if a better bit of kit would help me get clearer responses or if it makes no difference really.