Who’s There? Examining Voice Recognition with Deep Learning
Evan Sano, Alex Hall, A.L. Rooijen van, Taylor Togo · Scholarly Commons (University of the Pacific) · 2018
Is it possible to train a neural network to accurately detect the features of a human voice such as age, accent, and gender? If so, the answer could provide opportunities in fields such as biometrics, market research, security in identification, and Human-Computer Interaction. Recent developments in widely used technologies such as Google Home and Amazon’s Alexa use voice recognition to receive commands from their users. All of these technologies utilize various neural networks to detect voice commands. Being able to distinguish different voices could allow them to utilize a new layer of security in preventing unauthorized users to control the devices. It could also be used to collect data on users which the neural networks can detect. We believe that determining gender may not be as accurate due to the diversity in voice ranges; however height may be easier to determine. We can’t predict your future, but this trained computer will identify your age, accent, and gender solely from your voice. An interactive poster and computer combination awaits your experience at Voice Analysis. Participate in training the computer by inputting your own voice through a microphone! Come see the trained neural network work its magic towards human voice identification.