I don't know about whistling, but tone mis-detections were fairly common in the early days with some folks' voices, and they still arise in a few of the systems.
We tried to reduce this mis-detection within the hardware and software of a voice response system back in the early 1990s. This commonly involved intentionally requiring longer tone input detection before the tone action was triggered, but it was left programmable for the end-users.
I know someone that still triggering a tone detection in a conferencing system within the last few years; she'd end up muting herself in the call, which lead to appreciable frustration.
We tried to reduce this mis-detection within the hardware and software of a voice response system back in the early 1990s. This commonly involved intentionally requiring longer tone input detection before the tone action was triggered, but it was left programmable for the end-users.
I know someone that still triggering a tone detection in a conferencing system within the last few years; she'd end up muting herself in the call, which lead to appreciable frustration.