livekit-plugins-aws

What it is and what it does

This package bridges Amazon's AI services into LiveKit Agents, a framework for building voice-based AI applications. It provides four main components: a RealtimeModel for bidirectional speech-to-speech using Amazon Nova (with Nova 2 Sonic as the default), an LLM interface powered by Bedrock (defaulting to Nova 2 Lite), speech-to-text via Transcribe, and text-to-speech via Polly. The plugin handles the integration plumbing so you can compose these services into conversational agents.

The package is designed for developers building real-time voice agents that need AWS's AI capabilities. It supports multilingual conversations, automatic language detection and response mirroring, 18 expressive voices across multiple languages, and function calling for tool use. A recent breaking change switched the default realtime model from Nova Sonic 1.0 to Nova 2 Sonic, which adds text input support and mixed audio/text modalities; you can revert to the older model explicitly if needed.

Use it for:

• Build multilingual voice assistants that automatically detect and respond in the user's spoken language using Nova 2 Sonic.
• Create real-time speech-to-speech conversational agents with low-latency bidirectional streaming and interruption handling.
• Combine Transcribe + Nova LLM + Polly in a pipeline for speech-to-text, reasoning, and text-to-speech workflows.
• Add function calling and tool use to voice agents for weather lookups, web searches, or custom business logic.
• Deploy voice agents with multiple expressive voices and polyglot support across English, French, Italian, German, Spanish, Portuguese, and Hindi.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Integrates Amazon AWS AI services (Bedrock, Polly, Transcribe, Nova) into LiveKit Agents for speech-to-speech, text-to-speech, speech-to-text, and LLM capabilities.

Yes. The package is actively maintained (released 1 day ago), has no known vulnerabilities, low install friction, and Apache-2.0 licensing. Install it if you are building LiveKit voice agents and need AWS AI services (Bedrock, Transcribe, Polly, Nova). Ensure you have AWS credentials with Bedrock access and Python >=3.10.0 before proceeding.

Install

livekit-plugins-aws on PyPI

pip install livekit-plugins-aws

uv add livekit-plugins-aws

poetry add livekit-plugins-aws

Installing livekit-plugins-aws

Before you install

Low install friction; pure Python wheel with three runtime dependencies (aiobotocore, aws-sdk-transcribe-streaming, livekit-agents). Actively maintained with a release 1 day old and 13003 repository stars.

License in practice

Apache-2.0 permissive license allows commercial and private use with minimal restrictions; suitable for most projects.

Quickstart

pip install livekit-plugins-aws

from livekit.plugins import aws

model = aws.realtime.RealtimeModel.with_nova_sonic_2()
# or for LLM: llm = aws.llm.LLM()
# or for STT: stt = aws.stt.STT()
# or for TTS: tts = aws.tts.TTS()

Requires AWS credentials (AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_DEFAULT_REGION) with Bedrock access; Python >=3.10.0.

Verify before relying

• Whether Nova 2 Sonic realtime models require additional dependencies beyond the base install (description mentions optional [realtime] extra).
• Specific latency or throughput characteristics of realtime speech-to-speech streaming.
• Cost implications of using different Nova models and AWS services through this plugin.

Package facts

Evidence: livekit_plugins_aws-1.6.10-py3-none-any.whl

Keywords: ai, audio, aws, livekit, nova, realtime, sonic, video, voice

Tags

More Artificial Intelligence packages

Further reading