Voice AI & Speech Synthesis
Machines that listen, speak, and mean it
From acoustic models to neural codecs to real-time voice agents — the full speech stack, built for latency and trust.
Module
How Machines Speak
The TTS pipeline, audio as tokens, and controlling what the text doesn't say
Module
Hearing Machines
ASR at scale, the features underneath, and why real audio fights back
Module
Voices & Identity
Cloning, consent, and proving a voice is real
Module
Voice in Production
Latency budgets, speech-to-speech, and the full real-time agent stack
Capstone project
Conversational prototype
Design (and optionally prototype) a voice interaction: the turn-taking, error recovery, and persona. Record or script how it handles three tricky user inputs.
Deliverable: A dialogue spec or working demo showing graceful handling of interruptions and errors.
Milestones
Finish every lesson to earn the Voice AI & Speech Synthesis Certificate.