dB are a relative scale, so you need to know what it’s relative to for it to make the most sense.
For digital audio the reference value is usually “full scale” or the maximum loudness of any single sample.
This could be denoted as dB FS.
So, 0 is the biggest value and all other values are lower, hence they are negative numbers.
Sound levels, as per your conversation example, are typically measured relative to 20 micro pascals, which is roughly, the quietest sound humans can hear.
Most sounds are louder than that, so we get mostly positive numbers.
With digital audio recording, it is crucial to not clip the signal. Clipping is when the sound tries to be louder than 0dB, so the wave form gets held back at 0dB. Graphically, a sine wave has the top and bottom of its curve clipped off, hence the name.
As long as it doesn’t clip, you can in theory go as close as you like to 0dB.
In practice, recording engineers will usually aim for around -18dB, so that any unexpectedly loud signal does not clip.
Bear in mind though, that the recordings are often then processed/mixed and the finished version will often have a peaks of -0.1dB or 0dB.
Latest Answers