Choose an audio file and the waveform is drawn automatically once decoding finishes. Read the duration, sample rate, channels and peak level shown beneath it.
The vertical extent at each point is the range between the quietest and loudest sample in that slice of time. Tall sections are loud, flat sections are near silence, so you can spot dead air, clipping and song structure at a glance.
It is the loudest single sample expressed relative to digital full scale, where 0 dBFS is the maximum. A track peaking at -12 dBFS has plenty of headroom; one peaking at exactly 0.0 dBFS may already be clipped.
Yes, though the whole file is decoded into memory first, so hour-long recordings need a capable machine. The waveform itself is drawn from summarised peaks, so rendering stays fast once decoding finishes.