Live Meeting Assistant (LMA) with Amazon Transcribe, Amazon Bedrock, Amazon Quick Suite, and Strands Agents
Companion AWS blog post: Live Meeting Assistant with Amazon Transcribe, Amazon Bedrock, and Strands Agents
See CHANGELOG for latest features and fixes.
Questions?
You've likely experienced the challenge of taking notes during a meeting while trying to pay attention to the conversation. You've probably also experienced the need to quickly fact-check something that's been said, or look up information to answer a question that's just been asked in the call. Or maybe you have a team member that always joins meetings late, and expects you to send them a quick summary over chat to catch them up.
All of this, and more, is now possible with the Live Meeting Assistant (LMA).
Check out our original demo, and/or the updated demo to see the latest features including Virtual Participant, MCP based AI Agentic assistant - powered by Strands, and the new all voice assistant (think: Alexa or Siri in your meetings) - powered by Nova Sonic2, with optional Quick Suite integration.
LMABlogDemoV3.mp4
LMA-mcp-and-voice-assist-demo-sm.mp4
- Live transcription with speaker attribution — Powered by Amazon Transcribe with custom vocabulary and language model support
- Live translation — 75+ languages via Amazon Translate
- AI meeting assistant — Strands Agents SDK with Amazon Bedrock, with built-in tools for transcript search, web search, document search, and meeting history
- MCP server integration — Extend the assistant with external tools (Salesforce, Amazon Quick Suite, custom servers)
- On-demand and automatic summaries — Generate summaries, action items, and insights during and after meetings
- Virtual Participant — Headless Chrome bot joins Zoom, Teams, Chime, Google Meet, and WebEx meetings
- Voice assistant — Nova Sonic 2 or ElevenLabs voice responses with optional Simli animated avatar
- Meetings Query Tool — Semantic search across all past meeting transcripts via Bedrock Knowledge Base
- Embeddable components — iframe integration for embedding LMA in external applications
- Meeting recording — Optional stereo audio recordings stored in S3
- Meeting inventory — Searchable list of all meetings with sharing and access control
📖 Browse the docs site: aws-samples.github.io/amazon-transcribe-live-meeting-assistant
Full documentation source: docs/INDEX.md
Quick links:
- Prerequisites & Deployment
- Quick Start Guide
- Meeting Assistant
- Virtual Participant
- Voice Assistant
- MCP Servers
- Developer Guide
The LMA user starts a meeting session using the Stream Audio tab or Virtual Participant feature. A secure WebSocket connection streams two-channel audio to a Fargate-based WebSocket server, which relays it to Amazon Transcribe. Transcription results flow through Kinesis Data Streams to the Call Event Processor Lambda, which integrates with the Strands Agents SDK, Amazon Bedrock, and optionally Bedrock Knowledge Bases. Results are persisted to DynamoDB and pushed to the React web UI in real time via AppSync GraphQL subscriptions.
For full architecture details, see Infrastructure & Security.
| Region | Launch Stack |
|---|---|
| US East (N. Virginia) |  |
| US West (Oregon) | |
| AP Southeast (Sydney) | |
See Prerequisites & Deployment for full deployment instructions.
You are responsible for complying with legal, corporate, and ethical restrictions that apply to recording meetings and calls. Do not use this solution to stream, record, or transcribe calls if otherwise prohibited.
- Base infrastructure: ~$10/month (Fargate WebSocket server + VPC networking)
- VP EC2 instances: ~$33/month per warm instance
- Per-meeting usage: ~$0.17 per 5-minute call (varies by options)
See Troubleshooting for detailed cost breakdown and service pricing links.
Your contributions are always welcome! See CONTRIBUTING for more information.
See CONTRIBUTING for more information.
This sample code is made available under the MIT-0 license. See the LICENSE file.