Module 24
Voice & Multimodal
Real-time voice agents and multimodal generation.
- lessons
- 2
- steps
- 17
- time
- 26 min
01intermediate·14 min·9 steps
Real-Time Voice Agents
The reasoning is the easy part. Voice agents live or die on timing: when the caller stopped, how fast audio comes back, and what each minute costs.
- Compare chained, speech-to-speech and duplex architectures
- Why endpointing and barge-in decide how a call feels
- Read a per-turn latency budget
- Price a call across the three billing meters
Start →
02intermediate·12 min·8 steps
Multimodal Generation Pipelines
Generating images, video and audio is an asset pipeline of slow, expensive, non-reproducible jobs. Orchestrate it, prove provenance, and test it without exact matches.
- Treat a creative pipeline as a DAG of swappable stages
- Run long generation jobs asynchronously and safely
- Cut cost with draft-then-render and request caching
- Layer provenance: signed manifests plus watermarks
Start →
Read deeper
Guide chapters · Voice & Audio Agents
Next module · 25
Case Studies