AWS VoiceHands-Free Assistant

Natural voice interactions with Transcribe & Polly

Voice Command Center

Click to start demo

Conversation

AWS Transcribe

Real-time speech-to-text

AWS Polly

Natural text-to-speech

Voice Processing Pipeline

1. Continuous Transcription

AWS Transcribe converts all speech to text in real-time, highlighting commands after "Sous Chef".

transcribeStreaming.startStreamTranscription()

2. Wake Word Detection

When "Sous Chef" is detected, subsequent text is highlighted as a command for 10 seconds.

if (text.includes('sous chef')) { activateCommand() }

3. Command Processing

Highlighted commands are sent to Llama-4-Scout for recipe generation and guidance.

await llamaAPI.chat({command, context})

How It Works

Everything you say is transcribed. When you say "Sous Chef", the following words are highlighted as a command:

Regular: "Let me check what we have..."

Sous Chef, what can I make with tomatoes?

Regular: "I also have some basil..."

Sous Chef, add basil to the recipe