Image Alt

„Should Your Voice Deter­mine Whe­ther You Get Hired?“ – gele­sen auf

„Should Your Voice Deter­mine Whe­ther You Get Hired?“ – gele­sen auf

In lesen wir ein inter­es­san­tes Stück. Es beleuch­tet das Thema Stimme und Beruf und inwie­weit schon jetzt die Stimme über Eig­nung im Job ent­schei­det.

Tech­no­logy is chan­ging every facet of work, inclu­ding how com­pa­nies pro­file and select their employees. The deve­lop­ment of dif­fe­rent apps, soft­ware, and algo­rithms has pro­du­ced many novel metho­do­lo­gies for scree­ning job can­di­da­tes and eva­lua­ting their poten­tial fit for a role or orga­ni­za­tion.

The latest of such methods is voice pro­filing, the use of com­pu­ter-based algo­rithms to pre­dict job fit based on an ana­ly­sis of a candidate’s voice. Accord­ing to news reports, “regard­less of whe­ther you’re happy, sad, or cracking jokes, your voice has a hid­den, com­pli­ca­ted archi­tec­ture with an intrin­sic signature—much like a fin­ger­print. Through trial and error, the algo­rithms can get bet­ter at pre­dic­ting how things like energy and fun­da­men­tal fre­quency impact others—be they people watching a movie, or can­cer pati­ents cal­ling a help line.”

Alt­hough the idea that each voice is uni­que makes intui­tive sense, some voice pro­filing tools, such as Joba­line, are based on a rather uncon­ven­tio­nal pre­mise: Ins­tead of try­ing to decode a candidate’s per­so­na­lity, intel­li­gence, or mood state, they aim to pre­dict “the emo­tion that that voice is going to gene­rate on the lis­tener.” In other words, the algo­rithm func­tions as a mecha­ni­cal judge in a voice-based beauty con­test. Desi­ra­ble voices are invi­ted to the next round, where they are jud­ged by humans, while unde­s­i­ra­ble voices are eli­mi­na­ted from the con­test.

As with many other inno­va­tions in the space of HR tech­no­lo­gies, eva­lua­ting claims about the accu­racy of this method is dif­fi­cult until inde­pen­dent aca­de­mic rese­arch has been con­duc­ted. The good news, howe­ver, is that there is a well-estab­lished for­mula for tes­ting whe­ther the method works: (1) mea­sure fea­tures of can­di­da­tes’ voices, (2) mea­sure lis­teners’ reac­tions, and (3) mea­sure whe­ther lis­teners’ reac­tions relate to posi­tive orga­ni­za­tio­nal out­co­mes, such as reve­nues, pro­fits, and custo­mer satis­fac­tion. Then, cor­re­late 1 and 2, as well as 2 and 3. If a pat­tern is found, … im Ori­gi­nal wei­ter­le­sen.