While previous research has shown listeners' ability to determine age, ethnicity, and body size, sometimes based only on a single word, no known studies have explored the interaction of these variables in voice recognition.
To determine the factors involved, male speakers varying in ethnicity, height, weight, and age were asked to produce 5 monosyllable English words. These were analyzed for acoustic features (fundamental frequencies, formant frequencies, jitter & shimmer, voice onset time, etc.), then played for listeners asked to identify the four variables based only on that word.
Preliminary results show that listeners are most able to identify ethnicity and least accurate at identifying body size. Successful rating for Caucasian ethnicity was significantly higher than for other ethnicities. These results may have application in forensic linguistics in criminal identification.
Subscribe to:
Post Comments (Atom)
2 comments:
I'm taking a class from Dr. Baker and she talks about your study often. It sounds fascinating. I'm excited to see your final results. Could you tell us what conference this is for? Good job showing possible applications of your study. Your writing is concise, but perhaps too concise at times. Without giving way to excessive wordiness, experiment with adding details and power to your concise statements. Good job (and good luck!).
Congratulations on the progress you have made towards this Honors thesis. I hope that your presentation went well. I wish I could have finished the survey for your database, but I could not distinguish well enough. My responses would have been random guesses, and my schedule did not allow me to return even to guess!
Post a Comment