The Astro home robot from Amazon can manage video conversations, place grocery orders, and even get a drink. Its audio features make use of the company’s speech AI, Alexa, from Amazon. It was difficult for Astro to analyse sounds in real-time without becoming sidetracked by home noises or dogs. The sound waves travelling to the device’s microphone while you speak are reflected off of the walls and ceilings. For Alexa’s voice assistant, Astro, Amazon’s audio team has gathered some considerable brainpower to tackle a challenging issue. The team’s initial efforts were directed towards reducing background noise in audio and video conversations.
Related post: Brands are reluctant to try creative promotional ideas like Amazon Live
They employed a deep neural network (DNN), an AI-driven approach frequently used to solve audio and computer-vision issues. But they went one step further by creating a network architecture that dampens background noise as well as speech reverberation. An innovative approach was developed with the assistance of Athi Chhetri, an engineer in Amazon’s audio lab, to incorporate Astro’s speech-enhancement technology inside the voice-controlled house robot. Teams from Amazon laboratories all throughout the country collaborated to develop a cutting-edge solution. Numerous projects are currently in progress to further enhance the voice enhancement of Alexa and Astro.
Read more from Aboutamazon.com
