Voice Profiling Software Dehumanizes Hiring Decisions

Imagine that you’re a contestant in an audition round of The Voice, where you belt out your best “I Will Always Love You”. A minute passes. No reaction from the celebrity judges. You keep singing. Another minute, still no encouraging smile or nod. You strain to hit your highest note, pleading with your performance: “Please, please accept me! I am doing my best!” The song ends. No one wants you. Your family bow their heads in shame. Your mom cries. You stand on the stage, alone in the spotlight, heartbroken. A trap door opens beneath your feet and you slide screaming into Adam Levine’s basement torture maze.

Think that’s bad? In the real world, science has come up with something worse. A company called Jobaline offers “voice profiling” to predict job success based on how candidates sound; its algorithm identifies and analyzes over one thousand vocal characteristics by which it categorizes job applicants on suitability.

It’s horrible and dehumanizing, like all our other profiling (the racial kind is always a big hit!) Reliant on born-in, luck-of-the-genetic-draw factors that we can neither avoid or control. Regardless of mood or intent, according to NPR’s Aaarti Shehani, “your voice has a hidden, complicated architecture with an intrinsic signature – much like a fingerprint”.

This is not the only creepy algorithm system HR departments have been employing to help the company bottom line. Companies like Wal-Mart and Credit Suisse have been crunching data to predict which employees are “flight risks” who are likely to quit (easily remedied with a simple anklet attaching the worker to his or her cash register or cubicle) vs those deemed “sticky,” meaning in-it-for-the-long-haul. The information lets bosses either improve morale or get a head-start on a search for a replacement.

The inventors of such programs often enjoy the impeachable, amoral cloak of scientific legitimacy. When it comes to voice profiling, computers are not judging the speakers themselves; only the reactions the speaker’s voice provokes in other (presumably human) listeners…