Understanding where the audio begins

Good evening, friends.

Synchronizing video and audio, I came across a question about audio properties. I perceived the I can hear a noise or subsequent character speech before the the spectrogram in timeline.

Even when audio before the speech is a flat line, I can her a noise from the subsequent speech. I tried to cut the audio before the speech and I still get to hear the echo from the subsequent speech.

Now I wonder where the speech in fact begins: does the speech audio begin where the spectogram appoints or it beggins a little before?

When I cut the flat line of audio before the speech and moved it to a noiseless local, I still can hear a noise, though I can't hear the echo from the original speech it anteceded.

I'll show images in order to make you visualize what I am talking about.

[url=https://ibb.co/KXR2pnK][img]https://i.ibb.co/KXR2pnK/Audio-Timeline.png[/img][/url]

I can hear a bit of echo from the subsequent speech even in this flat line.

[url=https://ibb.co/NV1zPcf][img]https://i.ibb.co/NV1zPcf/Timeline.png[/img][/url]

I can hear the subsequent speech reverberation even when I cut the flat line.

And, when I moved this flat line to another point without any sound, I still can hear a little noise [but not the speech], though the spectrogram shows no sound at all.

What you think about?