The problem is more fundamental - the 'ping' sent has to be short in order to have good spatial definition (so distance
can be measured accurately), which means the bandwidth is very wide and spills into the audio spectrum. A typical
sender sends just 4 cycles of 40kHz so the power bandwidth is at least 10kHz wide and spills further than that. Since
the ping is very high power (to get adequate range), even a tiny fraction of signal can be well above the threshold of hearing.
However it may be that the circuit being used has a DC bias in the ping pulse which would then output a click across the
entire spectrum - I suspect some devices are better behaved than others.