We Can Identify A Friend’s Voice After Hearing Just Two Words

The human voice carries a great deal of information about who people are — information that can be decoded in a way that allows us to rapidly identify individuals and traits.

In fact, humans have such sophisticated vocal recognition abilities that we can outperform state-of-the-art machines designed for the same purpose.

In a new study, which was presented at the International Congress of Phonetic Sciences in Glasgow in August, linguists from the University of Montreal in Quebec found that more than 99 percent of the time, two words are enough for a person of normal hearing to distinguish the voice of a close friend or relative among other voices.

For the study, phoneticists created a “voice lineup,” which operated similarly to when the police line up a group of individuals with similar physical features before a witness. But in the voice lineup, several voices with similar acoustic properties were presented sequentially, with recordings of each voice uttering between one and 18 syllables.

A group of 44 French-speaking adult participants listened to a voice lineup featuring both the voice of a close friend or family member, and other similar-sounding voices.

The researchers found that the participants could identify the voices of their loved ones with 99.9 percent accuracy after hearing just four syllables — in this case, “merci beaucoup.” In contrast, machines have only a 92 percent chance of correctly identifying a voice.

“The auditory capacities of humans are exceptional in terms of identifying familiar voices,” Dr. Julien Plante-Hébert, the study’s lead author, said in a statement. “At birth, babies can already recognize the voice of their mothers and distinguish the sounds of foreign languages.”

This is because humans have a sophisticated sensory memory, which gives us the ability to record and later recall vast amounts of speech, including contextual information about what —> Read More