Ideas, in plain language
Engineering deep dives, research explained, and the occasional postmortem. Written by the people building Sunesis — not a content team.
Why emotion is the missing layer in AI
Voice and video give an AI a body. Emotion gives it a soul. Here's how we modeled warmth, hesitation, and excitement in synthesized speech.
We rewrote our inference stack. Here's what broke.
A postmortem on moving from per-request workers to a streaming inference graph, and the three bugs that almost shipped.
Designing the voice library: 200 voices, one catalog
How we organized, tagged, and made browsable a library of hundreds of voices without it feeling like a wall of names.
The science of emotional prosody
A deep dive into how humans vary pitch, pace, and pause to signal emotion — and how we teach a model to do the same.
We're opening our research lab to the public
Starting today, selected previews from our lab are available to any active account. Here's why, and how to opt in.
A pragmatic guide to voice cloning in production
Consent, licensing, quality thresholds, and the operational guardrails you need before shipping a cloned voice.
Never miss a post.
Subscribe to the RSS feed or follow our engineering updates.
SOC 2 Type II · GDPR · HIPAA-ready · No vendor lock-in