There's a lot of mythology floating around the ham radio and general EE community about the signal to noise ratio effects on various forms of modulation vs singing voice, such that sopranos have a natural SNR advantage on SSB and AM modulation but deep bass singer/speakers have a natural advantage over FM. Being primarily based on old wives tales I might have it backward and its interesting to contemplate the probably unstudied (at least in public journals?) effect over various cell phone and internet conferencing voice codecs.
There is a huge human factors problem missed in the article where are you trying to get attention or just provide a manufacturer CYA service. The first time an annunciator chimes out low altitude, that might be useful if the crew are not paying attention, but if the pilot and copilot are desperately trying to shout over the loud annunciator while they try to restart the engines its mostly just interfering with doing their job. Which has all kinds of smart phone UI implications, are you trying to bring something to the users attention like a ringtone, or merely be obnoxious (for ad purposes, I suppose?)
In the words of my loaner Hornet F7C:
Collision alert: front! Scanning! Collision alert: front!
Scanning! Scanning! Contact! Collision alert: Left! Contact!
Scanning! Sca- Contact! Collision alert: Front! Contact!