Conversational AI with Amazon Nova 2 Sonic

System prompts

Preset

Custom

Friendly Chat

Insurance Customer Service

Intelligent Assistant

Math Tutor

Model Capabilities

Voice

Tiffany

Matthew

Response timing

Medium

Fast Quick replies after brief pauses (~1.5s)

Medium Natural pacing for conversations (~1.75s)

Slow Longer pauses for thoughtful responses (~2s)

Output audio quality

24 kHz

8 kHz Low bandwidth, telephony quality

16 kHz Standard quality, balanced

24 kHz High quality audio output

Inference settings

Temperature

Top P

Response length

Low-latency mode

Prefetch customer data at auth — faster responses, less visible reasoning

Audio Settings

Initial buffer (ms)

Lower = faster response, higher = smoother playback

Tools

Enable or disable tools for the assistant to use during conversations.

Loading tools...

AWS Region

Asia Pacific (Tokyo)

US East (N. Virginia)

US West (Oregon)

Europe (Stockholm)

Configuration