Amazon Nova Sonic

Amazon Nova Sonic, the latest foundation model (FM) in Amazon Bedrock, simplifies speech-enabled app deployment

Amazon Nova Sonic blends speech interpretation and generation into a single model enabling developers to create natural and human-like conversational AI experiences with low latency and industry-leading cost

Amazon Nova Sonic to interface with external services and APIs and conduct operations in the customer's environment, including RAG knowledge grounding with enterprise data

Before utilizing Amazon Nova Sonic, toggle model access in the Amazon Bedrock console, just like other FMs. Enable the Amazon Nova Sonic for your account in the Model access area of the navigation pane under Amazon models

Amazon Bedrock's new bidirectional streaming API, Invoke Model With Bidirectional Stream, lets you develop HTTP/2-based, low-latency conversational experiences

Amazon Nova Sonic's new API can be accessed with model ID amazon.nova-sonic-v1:0

Amazon Nova Sonic prompts, prioritize conversational flow and intelligibility when heard rather than seen, and optimize the language for auditory comprehension

In American and British English dialects, the Amazon Nova Sonic can understand and produce speech in feminine and masculine voices

Amazon Nova Sonic blocks background noise and handles user interruptions without losing context