Voice Assistant

Voice Assistant Development

Overview
Building for Voice Assistants (Alexa, Google Assistant, Siri) is different from building web apps. It relies on Voice User Interfaces (VUI).

Architecture
1. Wake Word: "Alexa..." (On-device detection).
2. ASR (Automatic Speech Recognition): Converts audio to text ("Play jazz music").
3. NLU (Natural Language Understanding): Extracts Intent (`PlayMusic`) and Entity (`Genre: Jazz`).
4. Fulfillment: Lambda function executes logic (calls Spotify API).
5. TTS (Text to Speech): "Playing Jazz music."

Skills & Actions
- Alexa Skills Kit (ASK): Amazon's SDK.
- Actions on Google: Google's platform (now converging with Android Intents).

Design Challenges
- No Visuals: You can't show a list of 10 items. You must summarize ("Here are the top 3 results...").
- Discoverability: How does the user know what they can say? (Critical to provide help prompts).
- Ambiguity: "Play the rock" (Dwayne Johnson movie or Rock music?).

Current State
LLMs are replacing rigid NLU intents. "Alexa LLM" allows for dynamic, non-scripted conversations, moving away from the rigid command-and-control model.