Blog

Notes from the engine room

How we build self-hosted document and speech AI — and what we learn running it.

GuideJune 24, 20267 min read

Piper TTS in production: natural speech from your own servers

A practical guide to FastVision's speech stack: choosing Piper voices, streaming audio with sub-second first-byte, and shaping pronunciation with SSML.

Read the post

EngineeringJune 2, 20269 min read

Self-hosted OCR deep dive: PaddleOCR, DocTR and the preprocessing that actually matters

How FastVision's OCR pipeline squeezes accuracy out of hostile scans: OpenCV preprocessing, a two-engine strategy, and why bounding boxes are the real product.

Read the post

AnnouncementMay 12, 20265 min read

Introducing FastVision AI: OCR and TTS you actually own

Today we're launching FastVision AI 1.0 — a fully self-hosted OCR and Text-to-Speech platform. One API, your hardware, and not a single byte sent to a third-party AI service.

Read the post