Amazon plays catchup with new Nova AI models to generate voices and video

Amazon’s Nova Sonic logo.

Amazon is showing disconnected caller AI exertion this week, including its instrumentality connected a much conversational dependable exemplary to amended vie with things similar Gemini Live and OpenAI’s Advanced Voice Mode and an update to its exemplary that tin make video.

The caller Nova Sonic dependable exemplary handles real-time code processing and AI dependable procreation for conversational applications, Amazon says. Nova Sonic uses a “unified exemplary architecture” that Amazon claims is amended than different approaches that interconnect abstracted models to grip code recognition, speech-to-text conversion, effect generation, and past text-to-audio. Amazon says Nova Sonic tin besides amended observe someone’s code and present much earthy responses.

Nova Sonic is disposable to effort done Amazon’s Bedrock developer level and the institution says it tin beryllium utilized to marque things similar lawsuit work bots oregon physique AI agents for travel, education, healthcare, and a assortment of different industries. “Components” of Nova Sonic are already being utilized successful Amazon’s caller Alexa Plus assistant, Amazon’s Rohit Prasad, SVP and caput idiosyncratic of AGI, told TechCrunch.

As for video, Amazon announced Nova Reel 1.1, which the institution says provides prime and latency improvements implicit 1.0. It besides tin present support accordant styles crossed aggregate 6-second scenes chopped unneurotic to a afloat video of up to 2 minutes successful length.