"New Voices For The Voiceless: Synthetic Speech Gets An Upgrade"

Rupal Patel, a speech scientist at Northeastern University, has found a way to capture what she calls the "melody of the voice," and applied it to a striking new form of synthetic voice production. Patel's system works by taking samples of simple sounds produced by the individual whose voice she wants to synthesize—in her case, primarily children with severe speech disorders—sounds such as "aaahhhh," and combines these sounds with words spoken by a "healthy donor" of a similar age as the first individual. From the first set of sounds, they "can determine their pitch, the loudness, the breathiness of their voice, the changes in clarity," Patel says, and when combined with the "donor" voice, the new synthetic voice provides an uncanny approximation of a new voice for the speech impaired individual.

Read or listen to the complete story, "New Voices for the Voiceless: Synthetic Speech Gets and Upgrade" by Alix Spiegel here on NPR, and hear a few of Patel's synthetic voices!