#Latency
2 notas · última:
How I built a sub-500ms latency voice agent from scratch
The post argues that voice agents are an orchestration problem, and shows how a streaming pipeline with turn detection, cancellation, geography, and fast model choice cut latency from about 1.6s to around 400ms.
Everything You Know About Latency Is Wrong
The post argues that latency must be understood as a full distribution, not as averages or a few percentiles, and that coordinated omission makes many benchmarks and monitors lie about response time.