Taking new projects
AI · Voice Ordering for Restaurants
Voice POS
A restaurant POS that takes orders by voice: speak the table and the dishes, approve, done.
- Type
- AI system
- Year
- 2026
- Platforms
- Web · PWA

(01)Overview
A waiter holds the mic and says “Table twelve, two colas, one sea bass, a 20 cl rakı.” Speech is transcribed by Whisper running on the Apple GPU, a deterministic Turkish parser turns it into a table and line items, and an approval screen shows the interpreted order before it is added.
A frame from the live system

(02)The Problem
In a busy restaurant, tapping through menus on a POS is slow and error-prone. Voice is faster — but a POS cannot guess: a wrong table or a misheard dish costs real money.
(03)The Solution
Fast local speech recognition plus a parser that never improvises: exact, alias and fuzzy matching against the real menu, then a human confirms. Recognition runs on the host GPU because Docker on macOS cannot accelerate it — about 0.4 s instead of 11 s.
(04)What We Built
- 01
Whisper large-v3-turbo on MLX, kept alive as a macOS service
- 02
Table auto-detected from speech, or picked by tap
- 03
Turkish order parser with exact, alias and fuzzy matching — no AI tokens spent
- 04
Approval modal to review and edit before the order lands
- 05
Roles for admin, waiter and kitchen display
- 06
Demo transcript button so the whole flow works without a microphone
(05)Highlights
- 01
~0.4 s per utterance on the Apple GPU
- 02
No LLM in the parser — zero token cost
- 03
Approve-before-add safety step
(06)Outcome
- Speech-to-order in well under a second per sentence on local hardware
- Live demo at smartpos.mehditerzi.com
Our role
- Full Stack
- Speech AI
Tech stack
- Nuxt 3
- TypeScript
- Whisper large-v3-turbo
- MLX (Apple GPU)
- SQLite
- JWT roles
Related Projects
Something like this?
Want something like this? Every project can include design, development, deployment and ongoing improvements.


