RegaLabs / Blog

Notes from the build.

What we shipped, what we learned, and how the voice stack works under the hood.

4 min read

Deng Flash is live: streaming speech on every plan

Deng Flash renders speech as it streams — the first audio arrives while the rest is still being written. Here is what that changes for calls, previews, and live agents.

Read the post
6 min read

Provenance by default: why every render ships with a C2PA manifest

Synthetic speech is only trustworthy if you can check where it came from. A plain-language explainer of the C2PA manifest and the claim ledger behind every Rega render.

Read the post
7 min read

Kurdish-first is an engineering decision, not a slogan

What it actually takes to tune speech and recognition for Sorani, Badini, and Iraqi Arabic — diacritics, code-switching, and the text normalisation nobody warns you about.

Read the post