Sound travels at c₁ = 343 m/s in air. The lens is a biconvex shape (both faces bulge outward, radius of curvature R) made of a material where sound travels at a different speed c₂. At each face the wavefront bends exactly the way Snell's law says: sin θ₁ / c₁ = sin θ₂ / c₂ — the same relation that bends light at a glass surface, except here the "index" is a ratio of sound speeds, n = c₁/c₂, instead of light speeds.
When c₂ < c₁ (n > 1, e.g. dense gas or liquid), the biconvex shape converges the wave to a real focus in front of the lens, exactly like a converging optical lens — the same idea used to focus ultrasound in medical imaging. When c₂ > c₁ (n < 1, e.g. a lighter gas), the same biconvex shape instead diverges the wave; the rays and dashed line trace back to a virtual focus behind the source side.
The focal length follows the lensmaker's equation for a symmetric biconvex lens: 1/f = (n − 1)·(2/R), so f = R / (2·(n − 1)). Every ray drawn is individually refracted at both faces using the vector Snell's law — the "ray-traced" focal length is where the near-axis rays actually cross the axis, and it should track the lensmaker prediction closely for small apertures (and drift from it at the lens edge — real spherical aberration).