Practicing proper editing/cutting of audio samples improves heavily the skill you are looking for. Try for example to cut a speech (high quality, speaking your language) into its smallest parts. Try to separate the consonants from the vowels. This way you have to zoom into the waveform very deeply (see the describing shape of a particular sounding), until you get as small parts as small ones you would use in a wavetable synth.
While you are doing this, focus on the content, the sounding of a particular letter pronunced (which you are already familiar with, becaouse you speak the language), and the form it takes up. You will find out that you are getting familiar with the shapes and the corresponding soundings and after some hours of practice you will be able to tell the joining points in a particular word by eye and also recognize some letters only by seeing its form.
After cutting human speech, you can change to music or sound effects.
Another way would be to simply get a sounding similar to what you would like to achieve and have a look at it zoomed in.
I could imagine a practice like having a spectrally rich short sound and making heavy filterings on it and comparing the sounding differences with the visual differences. For example, when filtering with lowpass you can see that you loose the smaller bumps in the waveform, going the other way and doing highpass filtering will reduce bigger waves in the waveform.