This article announces the general availability of Amazon Nova 2 Sonic, an advanced speech-to-speech foundation model for building natural, real-time voice conversational AI applications.

• Delivers expressive, natural voices with improved conversational quality and human-like interactions
• Outperforms leading models on Big Bench Audio, BFCL, and ComplexFuncBench benchmarks
• Enhanced speech recognition handles alphanumeric inputs, accents, and background noise
• Expanded language support: Portuguese and Hindi added to existing languages
• Polyglot voices enable seamless code-switching between languages in single conversation
• Configurable turn-taking with adjustable voice activity detection sensitivity
• Crossmodal support allows switching between text and voice within same session
• Asynchronous tool calling enables multi-step tasks without pausing responses
• Direct integration with Amazon Connect, Vonage, Twilio, Audiocodes, LiveKit, Pipecat
• Available in US East, US West, Asia Pacific Tokyo, Europe Stockholm regions
• Maintains industry-leading price performance and low latency

Nova 2 Sonic enables developers to build sophisticated, globally-accessible voice applications with improved intelligence, expanded language capabilities, and seamless platform integrations.