Summary

  • Amazon has unveiled its new Amazon Nova Sonic voice AI, which it is offering to third-party developers to build real-time, conversational voice capabilities into their products using Amazon’s Bedrock web platform.
  • It combines three traditional separate models; speech recognition, language processing and speech synthesis into one, with the aim of making voice interactions more naturalistic.
  • It can handle two-way conversations and is able to understand when the user pauses, hesitates or interrupts and responds fluidly, while also integrating with other systems and proprietary tools.
  • Amazon said that it outperforms competitors such as OpenAI’s GPT-4o and Google’s Gemini Flash 2.0 on American and British English, and is nearly 80% cheaper.
  • It is now available to developers via Amazon Bedrock.

By Carl Franzen

Original Article