Observed arrival · 2026-10-06
The Trio: Haitian Creole, offline
The Trio presents three small models for recognizing Haitian Creole speech, answering questions, and speaking responses on a local device.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Haitian Creole speakers and offline voice-AI builders
- Worth noticing
- Fatiman is described as offering six voices from consenting speakers; the page also lists a 240 MB phone version.
Field notes
The page describes a three-stage pipeline: speech recognition, response generation, and speech synthesis, with the models running locally rather than relying on an internet connection. Its reported timing adds the three stages measured separately on a MacBook M3 Pro, so it is not presented as a simultaneous benchmark. Fatiman is listed at 1.3 GB, with a 240 MB phone version, and its six voices are attributed to consenting speakers.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue