On the other hand, according to classical physics the reflection coefficient is R = (Z1-Z2)2 / (Z1+Z2)2, where Z=ρc is acoustic impedance and Zair<<Zwater. That implies R≈1 due to the huge impedance mismatch.
But since the soapy water film is much thinner than it is wide, one has to take into account the kinematics of the film and air resistance on the film as it vibrates. For similar reasons a thick wood panel will reflect much more sound than a sheet of facial tissue will.
I remember an article about a recording studio (I forget where).
The idea was that studio setup would as best as possible capture instruments and vocals with minimal complication from sound reflection.
The solution was covering most of walls and ceiling with plastic foam bubbles.