Using the device and some generative AI tools, researchers were able to allow Ann to speak in near real time with a computerized voice trained on her own.