Skip to content
anythingai.space
Conversational AIintermediate12 min · 9 steps

Voice Assistants

Wrap a dialogue engine in speech: wake word, speech recognition, synthesis, voice-first design, screens, and recognising who is speaking.

audioactivatedtextreply textspeechUserWake wordon deviceASRspeech → textDialogue enginepipeline or LLMTTStext → speech
datasignalrequestresponse
1 / 9
Key ideaStep 1 of 9

A voice assistant is a chatbot with ears and a mouth

Voice assistants such as Siri, Alexa and Google Assistant put speech around a conversational system. ASR turns the user's speech into text. The dialogue engine (a classical pipeline or an LLM) decides the reply. TTS speaks the reply back.

A wake word starts each interaction, and voice activity detection tells real speech apart from background noise.

InsightThe middle box is the same dialogue engine you'd use for text. Voice adds a hard edge on each side: noisy audio in, natural speech out.
What changed
  • +Wake word
  • +ASR
  • +TTS

Lesson 6 of 8 in Conversational AI