Amazon is exhibiting off new AI expertise this week, together with its tackle a extra conversational voice mannequin to higher compete with issues like Gemini Stay and OpenAI’s Superior Voice Mode and an replace to its mannequin that may generate video.
The brand new Nova Sonic voice mannequin handles real-time speech processing and AI voice era for conversational purposes, Amazon says. Nova Sonic makes use of a “unified mannequin structure” that Amazon claims is healthier than different approaches that interconnect separate fashions to deal with speech recognition, speech-to-text conversion, response era, after which text-to-audio. Amazon says Nova Sonic can even higher detect somebody’s tone and ship extra pure responses.
Nova Sonic is accessible to attempt via Amazon’s Bedrock developer platform, and the corporate says it may be used to make issues like customer support bots or construct AI brokers for journey, schooling, healthcare, and a wide range of different industries. “Parts” of Nova Sonic are already being utilized in Amazon’s new Alexa Plus assistant, Amazon’s Rohit Prasad, SVP and head scientist of AGI, instructed Fintech.
As for video, Amazon introduced Nova Reel 1.1, which the corporate says gives high quality and latency enhancements over 1.0. It can also now maintain constant kinds throughout a number of six-second scenes lower collectively to create a full video of as much as two minutes in size.
