It depends on the sample format, but you can imagine a waveform graphic having a center line, the wave is sampled as many points progressing forward in time (towards the right) as sample values at the minimum (-1.0f), to values at the maximum (1.0f) and everything in between, (0.0 in the center). In this case, the samples represent the amplitude.
Again, depending on the sample format, those values could be floating 0.0 to 1.0, with 0.5 being the center; or 0 to 255, with 128 being the center, 0 to 65536, or -1.0...1.0 being the top.
That's called the "DEPTH". Some sample depths include ALLEGRO_AUDIO_DEPTH_INT16, ALLEGRO_AUDIO_DEPTH_INT8, etc. Here are the depth types: https://liballeg.org/a5docs/trunk/audio.html#allegro_audio_depth
Also, here's some code that renders an ALLEGRO_SAMPLE, which might be helpful: